# LEXUN — complete core-page text

> Generated from the shipped pages on every release; the short
> curated map is /llms.txt. Machine-readable registries:
> /system-manifest.json /model-registry.json /source-registry.json

## Homepage (https://lexun.co.uk/)

Stress-test a decision before you make it — LEXUN Skip to content LEX UN Decision Intelligence Simulate My decisions Evidence Teams Search Run a simulation Simulate Menu Close ✕ Decision engine Decisions Evidence Organisations Pricing Run a simulation Search pages, actions and your decisions The entrance shows a simulation computed in this browser. It has not run yet. Scroll The field is five thousand futures computed by this browser. With JavaScript off nothing is computed, so nothing is drawn — a picture is not shown in its place. LEXUN // Simulate the future Stress-test a decision before you make it. Run 5,000 reproducible scenarios on your own numbers. See the probability range, the downside, the breakpoint and the assumptions that matter — privately, in your browser. Describe the decision you are weighing up Run a free simulation → No account to run · No card · Simulation inputs stay on this device · Every result reproducible See a worked example → Runs the business-runway example below with its stated figures; nothing of yours is needed. For finance teams and advisers: book a Decision Audit · run an organisational pilot Business runway Answers: will cash stay above zero for the horizon — through a hire, a new site, a spend increase. Not: whether the plan is a good idea. Five figures · about two minutes Career change Answers: will savings carry you through the ramp — a new role, freelancing, retraining. Not: whether the move will make you happier or succeed. Five figures · about two minutes Major purchase Answers: what the commitment does to your buffer — a vehicle, a property, equipment. Not: whether it will hold its value. Six figures · about two minutes Or watch it refuse a question it cannot answer → Computed in this browser No tracking cookies Every result reproducible No invented figures Not financial advice Probability, not certainty. No accuracy % is claimed on any model until ten outcomes have resolved — next public score 17 September 2026 . The record . The answer one question, five thousand futures The answer, the breakpoint, and what moves them. Live · computed here, not fetched What you walk away with — a Decision Record Will this business keep positive cash for the next 12 months? The answer 60–70% JavaScript is off, so these are the worked example’s figures precomputed from the same inputs and seed (564), not a live run — the live demonstration replaces them when scripts are allowed. The same example, labelled step by step, is on the glass-box page . probability of positive cash Expected case Evidence quality moderate 3 of the 5 core evidence inputs are facts (horizon and optional fields not counted); 2 are assumptions. Key driver Monthly costs Plausible runway downside / expected / upside 9 / 12+ / 12+ mo Alternate futures Expected Downside Upside Stress The spread Open the full working the other three figures, the breakpoint, change one number, and the four shapes those futures take Most useful change to test Cut monthly costs — a 20% change swings survival by 62 pts What waiting 3 months could change −15 pts survival Growth needed for a 50/50 chance the growth at which survival is even 0.1%/mo Breakpoint The number where the future changes 0.1%/mo revenue growth Assumed 2%/mo Margin +1.9 pts of growth Above it — the future strengthens Below it — the future turns fragile Fragile Buffer you are here Change the future Monthly revenue — move it and the whole run again £9,000 £12,000 £15,000 The shapes those futures take Fact cash, revenue, costs Assumption growth, volatility Prediction the band above Freeze — required to save “Cash still positive on 31 August 2027.” A dated, checkable test — without one, a forecast is only a sketch. 5,000 paths · runway@1.0.0 · seed 564 · reproduce this exact result Your run ← back to the example — What moves it — and how you check it WORKED EXAMPLE · ONE INPUT AT A TIME · SAME SEED Re-running this simulation independently in your browser to check the receipt reproduces… See the technical receipt model runway@1.0.0 · seed 564 (input-derived) · paths 5,000 · reproducibility_id rw_52f9a7c3_234 And it refuses to guess. Ask it something the released models cannot answer and it says Insufficient evidence rather than producing a number. Confidence for this run is stated in the facts above. Reproduce this result → · Verify a sealed record → How it works — the method, in full Every input is classified before anything is computed, and the classification is shown rather than assumed. Facts are numbers you gave us, assumptions are numbers you or we estimated, and the prediction is the only thing the engine adds. Live engine inputs, worked example Input Value Classified as Cash today £80,000 FACT Monthly revenue £12,000 FACT Monthly costs £18,000 FACT Revenue growth 2%/mo ASSUMPTION Volatility 15%/mo ASSUMPTION Survival band 60–70% PREDICTION The simulation is a seeded Monte-Carlo run: 5,000 monthly cash paths in antithetic pairs, revenue following a log-normal step about your stated growth rate — the standard model for a quantity that moves in proportion to its own size and cannot go below zero. The seed is derived from your inputs, so the same inputs always produce the same answer — that is what makes the receipt above checkable rather than decorative. Reference class: 38.4% of UK businesses born 2019 survived five years — ONS Business Demography 2024 . Bundled snapshot, labelled as such — not live. Full methodology, including what this model is bad at → The Decision Reality Loop the loop LEXUN is built around Every prediction stays open until reality closes it. Frozen before the answer is known. Scored when reality arrives. Misses published at the same size as the hits. Predict → Freeze → Monitor → Resolve → Score → Recalibrate The public record, today We publish the miss f/003 · Brier 0.25 “July 2026 CPI, as first published, lower than June’s 2.6%?” Band 42–58% — deliberately uncertain — and reality answered up : 2.9%. The entry stays on the record at the same size as the hits. Editorial forecasts resolved 6 of 12 issued (2 open, 4 withdrawn) against first-published figures — these test our forecasting discipline, not the decision models, which have 0 resolved outcomes Open forecasts 2 next: 2026-09-17 Calibration table Awaiting data Published the moment the sample is honest. Model versions runway 1.0 · career 1.1 · major purchase 1.0 Every result carries its version. Editorial forecasts test forecasting discipline, not the decision models — each model earns its own calibration from resolved outcomes. Methodology → · Forecast register → Your own record, on this device 0 frozen · 0 resolved · 0 ready to score A different kind of evidence — and it is not accuracy 711 automated engine checks , all passing: determinism, evidence gates, tamper detection, scoring arithmetic. They show the engine behaves as specified — not that a forecast will come true. It does not prove a forecast came true, and nothing here will use it as if it did. What the checks cover → Three fair objections, answered Why not just ask an AI chatbot? A chatbot’s number has no seed, no receipt, no frozen criterion and no scored history — LEXUN’s has all four, and answers fewer questions because of it. The comparison, point by point → What would prove this system wrong: a register entry edited after issue · a resolution scored against a source not named in advance · an accuracy claim before ten resolved outcomes in a domain. Each is checkable from the raw ledger — the full falsification list → Anyone can copy this interface in a weekend. A public scoring history becomes harder to replicate as verified outcomes accumulate. Why that is the whole defence → What LEXUN can model today Three released models. Everything else is refused, honestly. The full catalogue, weaknesses stated, lives on the templates page . Business cash · Career change · Major purchase — each opens its worked example. Free · £0 The full engine Three models, receipts, freeze and score, in your browser. The maths never gets better when you pay. Run a decision → Pro · £9/mo or £79/yr Keep the record Every decision kept, exports, PDF reports and the calibration view. A 14-day trial, no card. Checkout appears only once Stripe is switched on, and the page says so. See plans → Organisations · £49/seat The Decision Record, for teams £49 a seat a month (£490 a year), Enterprise after a pilot. Everything not yet built is labelled planned, not sold. Decisions for Organisations → Who we are Founder-built and independent — no advertisers, no data brokers, no outside investors A trading name of CA Capital Limited; not connected to any cryptocurrency exchange or similarly-named company About LEXUN → · Trust & validation → · Companies House record → · Contact → Private by architecture, not by promise Decisions are computed in your browser and stay there — no server copy No trackers, no cross-site cookies, no analytics on decision content One stated exception: the optional Copilot sends what you type to it — and, when you ask it to explain a result, that result’s band and percentages, never your figures — to Anthropic’s API. Saved decisions never go UK GDPR and the Data Protection Act 2018 · an estimate for you to use, never an automated decision about you Privacy & your rights → · Cookies → · Security → · Terms → Run a simulation → Free, in your browser. No account, no card. 01 The answer 02 Reality loop 03 Start Product Simulator My decisions Evidence For teams Pricing Company About LEXUN Contact Roadmap Resources API Status Changelog Legal Privacy notice Terms of service Cookies and local storage Security The full site — every page LEXUN publishes Home Runway guide Career-change guide Big-purchase guide Glass box vs black box Reproduce a result Glossary How LEXUN compares Why this is hard to copy Careers For education API System status Changelog Cite & check our data Country coverage Labs: is my role exposed to AI? © 2026 LEXUN — Decision Intelligence. Decision intelligence, not financial, legal or medical advice. Operated by CA Capital Limited, registered in England and Wales, company number 10848369, registered office 320 Firecrest Court, Centre Park, Warrington, United Kingdom, WA1 1RG. ICO registration ZC183768. View the record at Companies House . Contact: hello@lexun.co.uk No tracking cookies. Local storage is used only for features you choose. Details Understood Run a simulation →

## About (https://lexun.co.uk/about)

About LEXUN — who builds it and why Skip to content LEX UN Simulate My decisions Evidence Teams Get your Key Search Ctrl K Menu Run a simulation About Built in the UK, in public, against the grain. Most prediction products claim spectacular accuracy and publish nothing you can check. LEXUN is built on the opposite bet: that the way to earn trust in a forecast is to show the method, freeze the prediction, and score it against what actually happens — in public. On this page Expand all The story Who is behind it Legal entity What we believe Contact The story LEXUN started in 2026 as a single question: why does every "AI decision" tool give you a confident answer and no way to find out if it was right? The founder built the first release-gated model — UK business cash runway — around three rules that still govern every line of the product: no displayed number without a source, "insufficient evidence" is a real answer, and every forecast freezes its resolution criterion so success cannot be redefined after the event. Who is behind it Founder & product — LEXUN is currently founder-built and founder-run, from the United Kingdom. We do not pad this page with stock photos or invented advisers: as the team grows, real names, roles and photographs will be published here. That is a deliberately uncomfortable honesty — the same standard we apply to our accuracy numbers. Legal entity LEXUN is a service of CA Capital Limited , a company registered at Companies House in the United Kingdom. It is registered in England and Wales, company number 10848369, registered office 320 Firecrest Court, Centre Park, Warrington, United Kingdom, WA1 1RG. ICO registration ZC183768. That record is public and checkable — view it at Companies House rather than taking this page’s word for it. LEXUN is a trading name of CA Capital Limited and is not connected to any cryptocurrency exchange or to any other company using a similar name. Correspondence: hello@lexun.co.uk . What we believe Calibration beats confidence. A 65% forecast that is right 65% of the time is worth more than a "99% accurate" claim nobody can audit. Reproducibility is respect. Every result carries its seed, model version and input hash — anyone can re-run it and get the identical output. Try it . The small number is published. Six outcomes have resolved so far, and our Accuracy Centre says exactly that — and it shows no accuracy percentage until at least ten have. The numbers will mean something precisely because we refused to invent them early. The record is the moat, not the screen. The interface is copyable in a weekend; a scored track record can only be filled in forwards. We set out what that defence is — and what it is not — on the defensibility page . Contact General: hello@lexun.co.uk · Privacy: privacy@lexun.co.uk Where to go next Why this is hard to copy The accuracy record is the asset a competitor cannot rewrite. Accuracy Centre The methodology, the sources, and what we do not yet know. Contact us Ask something directly, and get a person. Questions people actually ask What is LEXUN? LEXUN is a decision-intelligence engine that runs in your browser. It turns a consequential business or personal decision into a banded probability with the evidence, the assumptions and the strongest counter-argument shown alongside it, then freezes the forecast and scores it against reality once the outcome is known. Can LEXUN predict the future? No, and it never claims to. LEXUN estimates possible futures: it runs 5,000 seeded simulation paths over the figures you enter, shows a probability band rather than a false point, states what would change the answer, and records the forecast so reality can grade it later. Certainty is the one thing it refuses to sell. How accurate is LEXUN? The register publishes no accuracy percentage for any domain until at least ten outcomes have resolved there, because a smaller sample cannot support the claim. What exists today is public and checkable: twelve forecasts issued, six resolved against official figures as first published, each Brier-scored on the forecast register. Is LEXUN free? Yes. Running a decision needs no account and no card. Saving your record needs only a passcode that encrypts it on your own device. Is my data private? Your decision data stays in your browser, encrypted under your own passcode. The one exception is the optional Copilot: it sends what you deliberately type into it to the AI provider (if you type a figure there, that figure is included), and the result tools send only the filtered result summary that the Privacy notice describes. There is no server-side store of scenarios and no advertising or tracking cookies. What can leave the device is only what you choose to send: an email address on a form, a payment made on Stripe's own page, and the words you type to the optional Copilot (which never receives your saved decisions or the figures you entered). Product Simulator My decisions Evidence For teams Pricing Company About LEXUN Contact Roadmap Resources API Status Changelog Legal Privacy notice Terms of service Cookies and local storage Security The full site — every page LEXUN publishes Home Runway guide Career-change guide Big-purchase guide Glass box vs black box Reproduce a result Glossary How LEXUN compares Why this is hard to copy Careers For education API System status Changelog Cite & check our data Country coverage Labs: is my role exposed to AI? © 2026 LEXUN — Decision Intelligence. Decision intelligence, not financial, legal or medical advice. Operated by CA Capital Limited, registered in England and Wales, company number 10848369, registered office 320 Firecrest Court, Centre Park, Warrington, United Kingdom, WA1 1RG. View the record at Companies House . Contact: hello@lexun.co.uk

## Guide: business cash runway (https://lexun.co.uk/learn-runway)

Cash runway probability — will it survive? — LEXUN Skip to content LEX UN Simulate My decisions Evidence Teams Get your Key Search Ctrl K Menu Run a simulation Guide · Business cash runway How long will my business cash last? Last updated 7 August 2026 · written against runway@1.0.0 · by the LEXUN team at CA Capital Limited The checkable facts: 5,000 simulated paths per run · free, no account or card · your figures stay on this device (the optional Copilot is the one exception, and it sends only what you deliberately type or ask it to summarise) · forecasts frozen before outcomes are known, then Brier-scored in public — misses included. The formula everyone uses flatters you, because real revenue doesn't arrive in a straight line. Here is the honest way to work out your runway — and a simulation you can reproduce yourself in two minutes. On this page Expand all The straight-line answer (and why it lies) What survival probability actually looks like The three numbers that move the answer most The environment you're operating in Work out yours Method note — maths, assumptions, failure modes The straight-line answer (and why it lies) The formula in every textbook is runway = cash ÷ (monthly costs − monthly revenue) . If you have £80,000, spend £18,000 a month and bring in £12,000, you're burning £6,000 a month, so the answer is "about 13 months". Simple. The problem: that calculation assumes next month's revenue equals this month's, every month, forever. Small-business revenue rarely swings less than 10–15% month to month. Volatility is not noise that averages out — a bad quarter early can end the business even when the averages say you were fine . Straight-line runway is the best case wearing a disguise. What survival probability actually looks like The honest question isn't "how many months of average burn do I have?" but "what fraction of the plausible futures keep my cash above zero for the horizon I care about?" That needs a simulation: thousands of possible revenue paths with your stated growth and volatility, counting how many survive. A real worked example (run on this site's engine — deterministic, so you'll get the identical result): cash £80,000 · revenue £12,000/mo · costs £18,000/mo · growth 2%/mo · volatility 15%/mo · horizon 12m Straight-line says ~13 months. The simulation says the survival probability over 12 months is 60–70% — roughly one in three plausible futures runs out of cash — and the plausible runway range is 9 to 12+ months (12+ meaning the path was still solvent when the 12-month window closed — month 13 was never simulated). Same numbers, very different decision. Reproduce this result yourself → The three numbers that move the answer most Monthly costs. In the example above, ±20% on costs swings survival by more than any other input. Costs are also the number you control most directly — which is why cost discipline beats revenue optimism in a cash crisis. Revenue volatility. Two businesses with identical averages and different volatility have very different survival odds. If one customer is more than ~20% of revenue, your true volatility is lumpier than your spreadsheet shows. Growth assumptions. Compound monthly growth above ~5% is very hard to sustain for a year. If your survival verdict flips when you halve your growth assumption, the decision is resting on hope. The environment you're operating in For context, not destiny: 38.4% of UK businesses born in 2019 were still trading five years later ( ONS Business Demography 2024 ). Roughly three in five don't make it to year five — which is exactly why watching your cash honestly matters more than feeling optimistic about it. What no calculator can tell you: whether a specific customer will churn, whether that invoice will pay on time, or whether you'll land the contract. A good runway model doesn't predict these — it shows you how exposed you are if they go wrong, and refuses to answer at all when the inputs are guesses. LEXUN's engine does exactly that: it declines to run on unknowns rather than inventing values, and it shows every result as an honest range, never a false point. Work out yours The LEXUN runway analysis runs entirely in your browser — your figures stay on this device; the optional Copilot is the one exception, and it sends only what you deliberately type or ask it to summarise. Enter your cash, revenue, costs and volatility, mark which are facts and which are assumptions, and get a survival band, the drivers that move it, a stress scenario, and a frozen criterion so you can come back and check whether we were right. Run the runway analysis → Method note — maths, assumptions, failure modes The maths. Cash evolves monthly: revenue follows a log-normal multiplicative step, rev t = rev t−1 × exp(ln(1 + g) − σ²/2 + σ·ε) with ε ~ N(0,1); costs grow at a stated rate, committed one-off flows land in their month, and a path ends when cash ≤ 0. The −σ²/2 term makes the expected monthly multiplier exactly 1 + g at every volatility, so raising volatility widens the spread of outcomes rather than adding growth to them, and the step is strictly positive, so no single month can set revenue to zero. (It could before: the earlier form was × (1 + g + σ·ε) floored at zero, and at the 80%-a-month volatility this site allows, 64% of paths had revenue destroyed by arithmetic rather than by trading — see the changelog .) 5,000 seeded paths (mulberry32 PRNG, Box–Muller normals) run in antithetic pairs — each block of monthly shocks used once as drawn and once negated, which cuts the Monte-Carlo error by about a third for the same work — give P(survival) with a binomial standard error that is displayed, never hidden. Bands are the honest output, not a stylistic choice. Key assumptions. Monthly shocks are independent and normal (real revenue is often autocorrelated and fat-tailed — the stress scenario partly compensates); costs are deterministic apart from the stated growth; no funding events are modelled — a raise, loan or large invoice must be entered as a committed inflow. Failure modes. Pre-revenue firms with lumpy invoicing violate the smooth-revenue assumption; volatility under ~10%/mo is rarely true for small firms and will flatter the result; growth beyond ±50%/mo or volatility beyond 80%/mo is refused as outside model validity. When most paths survive the horizon, runway percentiles read "12+" — the model does not extrapolate beyond the horizon you set. References. Survival framing: ONS Business Demography 2024 (five-year survival 38.4%, environment context only). Scoring: Brier (1950); calibration methodology on the Accuracy Centre . Unfamiliar terms — survival band, frozen criterion, Brier score, calibration — are defined in the decision glossary . Related guides. The same method applied to two other decisions: leaving a job , and a large purchase or commitment . To satisfy yourself the numbers are real before you rely on any of them, re-run a published result yourself and check it against the calibration record . Stated weaknesses — what this model cannot see Every model is a simplification, and a model that will not name its own is asking for trust it has not earned. These are runway@1.0.0's, stated from the recurrence itself, not from marketing: One revenue stream, one volatility. A single monthly shock scales your whole revenue. Losing one large client — a discrete cliff, not a wobble — is exactly the event this shape cannot represent, and for many small firms it is the event that matters. Months don't remember each other. Shocks are independent month to month. Real downturns cluster: a bad quarter makes the next month worse, and this model does not know that. Costs never shock. Costs compound smoothly at the growth rate you give. A lawsuit, a rent review, an emergency hire — the model has nowhere for them to land. Committed amounts land in full, on the booked month. A funding round that slips six weeks does not slip here. Insolvency is cash at or below £0, full stop. No overdraft headroom, no invoice factoring, no emergency bridge. Real businesses have messier endings, in both directions. No seasonality, no VAT or tax timing. Sensitivity probes shift each driver ±20%. Near a cliff edge every driver saturates at ±100 pts — a true statement about your situation, but not a ranking anymore. If your decision hinges on one of these blind spots, this model is the wrong tool, and the honest move is to say so — which is what the insufficient-evidence state exists for. Where to go next Run your own numbers The worked example, with your figures instead. Career-change guide If the honest answer is a change of income. Ask LEXUN Say it in your own words instead of filling in a form. Questions people actually ask How do I calculate my business runway? Divide the cash you actually hold by your true monthly net burn — all cash out minus all cash in. That gives a single deterministic number; LEXUN goes further by simulating 5,000 possible paths for your revenue and costs, because real months vary, and reports the probability your cash stays above zero for the whole horizon rather than one falsely precise date. What is a good cash runway for a small business? There is no universal figure, and anyone quoting one is guessing. What matters is the probability your cash survives your own horizon given your real revenue volatility — a business with steady contracted income can safely run shorter than one with lumpy seasonal sales. LEXUN computes that probability from your figures instead of applying a rule of thumb. Does LEXUN use my bank data to work this out? No. You type your figures, or import a CSV that is read entirely in your browser — nothing is uploaded, and the release gate literally verifies that no network request leaves the page during import. An imported number still has to be classified by you before it counts as a fact. Product Simulator My decisions Evidence For teams Pricing Company About LEXUN Contact Roadmap Resources API Status Changelog Legal Privacy notice Terms of service Cookies and local storage Security The full site — every page LEXUN publishes Home Runway guide Career-change guide Big-purchase guide Glass box vs black box Reproduce a result Glossary How LEXUN compares Why this is hard to copy Careers For education API System status Changelog Cite & check our data Country coverage Labs: is my role exposed to AI? © 2026 LEXUN — Decision Intelligence. Decision intelligence, not financial, legal or medical advice. Operated by CA Capital Limited, registered in England and Wales, company number 10848369, registered office 320 Firecrest Court, Centre Park, Warrington, United Kingdom, WA1 1RG. ICO registration ZC183768. View the record at Companies House . Contact: hello@lexun.co.uk

## Guide: career change affordability (https://lexun.co.uk/learn-career)

Career change calculator — can I afford to quit? — LEXUN Skip to content LEX UN Simulate My decisions Evidence Teams Get your Key Search Ctrl K Menu Run a simulation Guide · Career change Can I afford to change careers? Last updated 7 August 2026 · written against career@1.1.0 · by the LEXUN team at CA Capital Limited The checkable facts: 5,000 simulated paths per run · free, no account or card · your figures stay on this device (the optional Copilot is the one exception, and it sends only what you deliberately type or ask it to summarise) · forecasts frozen before outcomes are known, then Brier-scored in public — misses included. Whether the move is right is a life question no calculator should pretend to answer. Whether you can afford it is a maths question — and it has four inputs. On this page Expand all The only four numbers that matter financially A real worked example What people get wrong The honest context Work out yours Method note — maths, assumptions, failure modes The only four numbers that matter financially Your buffer — savings you could genuinely spend during the transition (not your ISA-you'd-never-touch). Your essential monthly costs — from bank statements, not memory. Most people's true minimum month is 10–20% higher than their guess. The income ramp — how many months until the new path pays what you expect. A salaried job ramps in one month; freelancing and founding typically take six or more. Income volatility — salaried income barely moves; freelance income routinely swings 20–40% month to month, which drains buffers faster than the average suggests. The question a simulation can answer honestly: in what fraction of plausible futures does your buffer stay above zero until the new income carries you? A real worked example Someone leaving a £2,600/month job to freelance, with £12,000 saved, £2,400/month essential costs, expecting £2,600/month after a 6-month ramp, at 30% monthly volatility: buffer £12,000 · essentials £2,400/mo · target income £2,600/mo · ramp 6mo · volatility 30% · horizon 12m On these stated inputs the engine says 90–100% — financially resilient . But the stress scenario (income −30%, ramp 3 months slower) collapses to 7% survival . That gap is the real finding: the plan works if the income estimate is honest , and fails badly if it's hope. The single most valuable thing this person can do before resigning is turn that income assumption into evidence — a written offer, or two or three signed clients. Reproduce this result yourself → What people get wrong Counting the average, not the ramp. Most failed transitions die in months 2–5, while income is still ramping — not at the average income level. Optimistic essential costs. If you've never had a bank-statement month at your claimed minimum, it isn't your minimum. Fast ramp + volatile income. Income that arrives immediately at full rate *and* swings 30% a month is a contradiction. Pick one honestly. Forgetting the return option. The ability to go back to employment mid-way makes your true downside less bad than any simulation shows — in your favour, and worth stating. The honest context Career moves are normal, not reckless: roughly 2.9 million UK workers changed jobs in 2025 ( Indeed Hiring Lab, citing ONS ). What we won't tell you: there is no defensible statistic for the probability a career change "succeeds" — definitions vary too much to pool, and anyone quoting one at you is making it up. LEXUN's career model therefore forecasts only the resolvable question — does your buffer survive the transition? — and says so plainly rather than inventing a success rate. Work out yours Runs entirely in your browser; your figures stay on this device — the optional Copilot is the one exception, and it sends only what you deliberately type or ask it to summarise. You'll get a survival band, the break-even income where the odds flip, what waiting-and-saving another few months does, and a frozen criterion to come back and score against reality. Run the career-change analysis → Method note — maths, assumptions, failure modes The maths. The model simulates your financial buffer through an income transition: essential costs drain the buffer monthly while income ramps linearly from zero (or your current income end-date) to the expected new level over the stated ramp months, with log-normal shocks of mean exactly one scaled by income volatility — so raising volatility widens the spread without moving the average, and a month’s income can be small but never negative. 5,000 seeded paths; a path fails when the buffer hits zero. Output is P(buffer survives the horizon) as a band. Key assumptions. The ramp is linear — real freelance ramps are lumpy; essential costs are treated as fixed (inflation and lifestyle drift are not modelled); the model deliberately refuses to forecast career "success", which has no defensible base rate — only the resolvable buffer question. Failure modes. Salaried moves with a signed offer make most inputs facts and the band tight; founder/freelance moves rest on assumed income, and the confidence rating will say so. Partner income or emergency borrowing are not modelled — if they exist, your true downside is better than shown. References. Labour-mobility context: ~2.9M UK job changers in 2025 (Indeed Hiring Lab / ONS). An independent deterministic heuristic cross-checks every simulation and material disagreement is displayed, not hidden — details on the Accuracy Centre . Related guides. The same method applied to two other decisions: business or personal runway , and a large purchase or commitment . Terms such as survival band, frozen criterion and Brier score are defined in the decision glossary , and you can re-run a published result yourself before relying on any of it. Stated weaknesses — what this model cannot see These are career@1.1.0's blind spots, stated from the recurrence itself: It scores buffer survival, and only that. Whether the career change succeeds has no defensible base rate, so the model refuses to score it rather than inventing one. What you get is the financial floor under the decision, not the decision. The ramp is a straight line. Income climbs linearly to full over the ramp months. Real ramps are lumpy — probation, first-invoice delays, commission cliffs — and a lump in the wrong month can matter more than the slope. Essential costs are frozen for the whole horizon. No rent rise, no cost shock, no life event. Income shocks are symmetric and independent month to month. A freelancer's feast-and-famine autocorrelation is not represented. Severance, benefits and side income don't exist unless you fold them into the numbers you give it. If one of these is the crux of your situation, the model is the wrong tool — the insufficient-evidence state is how it says so when it can tell, and this list is how we say so when it cannot. Where to go next Run your own numbers The worked example, with your figures instead. The public record Every forecast, scored against what happened. Runway guide How long your savings cover a gap between jobs. Questions people actually ask Can I afford to change careers? That is a resolvable question, and it is the one LEXUN's career model actually answers: given your buffer, essential costs and the income gap of the transition, what is the probability your finances survive the switch? It deliberately does not score whether the new career will succeed — no defensible base rate exists for that, and the model says so on every result. How much money should I save before quitting my job? Enough that your buffer survives your realistic transition window at a probability you can live with — which depends on your essential costs, any partner income and how long re-employment takes, not on a universal number of months. LEXUN simulates 5,000 paths over your actual figures and shows the income level at which the odds reach 50/50. Will AI take my job? Nobody can give you a single honest percentage for that, and LEXUN refuses to invent one. Its automation-exposure checker is deliberately qualitative — it helps you reason about which tasks in your role are exposed, and it is labelled as reasoning support, never as a prediction. Product Simulator My decisions Evidence For teams Pricing Company About LEXUN Contact Roadmap Resources API Status Changelog Legal Privacy notice Terms of service Cookies and local storage Security The full site — every page LEXUN publishes Home Runway guide Career-change guide Big-purchase guide Glass box vs black box Reproduce a result Glossary How LEXUN compares Why this is hard to copy Careers For education API System status Changelog Cite & check our data Country coverage Labs: is my role exposed to AI? © 2026 LEXUN — Decision Intelligence. Decision intelligence, not financial, legal or medical advice. Operated by CA Capital Limited, registered in England and Wales, company number 10848369, registered office 320 Firecrest Court, Centre Park, Warrington, United Kingdom, WA1 1RG. ICO registration ZC183768. View the record at Companies House . Contact: hello@lexun.co.uk

## Guide: big-purchase affordability (https://lexun.co.uk/learn-purchase)

Can I afford it? Big-purchase stress test — LEXUN Skip to content LEX UN Simulate My decisions Evidence Teams Get your Key Search Ctrl K Menu Run a simulation Guide · Major purchase Can I afford this big purchase? Last updated 7 August 2026 · written against opportunity@1.0.0 · by the LEXUN team at CA Capital Limited The checkable facts: 5,000 simulated paths per run · free, no account or card · your figures stay on this device (the optional Copilot is the one exception, and it sends only what you deliberately type or ask it to summarise) · forecasts frozen before outcomes are known, then Brier-scored in public — misses included. "Can I cover the monthly payment?" is the wrong question. The right one: after the upfront cost leaves my savings, does my buffer survive the real running costs for the next year? On this page Expand all Why "I can cover the payment" fails A real worked example What people get wrong The honest context Work out yours Method note — maths, assumptions, failure modes Why "I can cover the payment" fails Big commitments break budgets through two mechanisms the advertised price hides. First, the upfront cost drains the cushion that was protecting you from everything else — the boiler, the dental bill, the month between jobs. Second, the real monthly cost is almost never the advertised monthly cost : insurance, maintenance, service charges and repairs routinely add 15–30% to the number on the advert. The honest test is buffer survival: simulate your income (with its real month-to-month swings) against your essential costs plus the full commitment, starting from the buffer you have after the upfront payment leaves, and count what fraction of plausible futures keep you above zero. A real worked example Someone on £2,600/month with £12,000 saved and £2,050/month essential costs, buying a £6,000 car with £550/month total running costs (finance + insurance + fuel + maintenance): buffer £12,000 · upfront £6,000 · commitment £550/mo · income £2,600/mo · essentials £2,050/mo · volatility 8% · horizon 12m On these stated inputs: 95–100% — comfortably affordable . But the stress scenario (running costs +25%, income −15%) drops to 41% — near a coin flip. The verdict isn't "yes, buy it"; it's "affordable if the £550 is real ". Get the insurance quote and a realistic maintenance number in writing before trusting the green light. The engine also reports the largest commitment this budget can carry with the odds still on your side: about £1,050/month — useful when comparing options. Reproduce this result yourself → What people get wrong Using the advertised monthly figure. The advert shows the finance payment; you'll pay finance + insurance + running costs + the repair in month seven. Ignoring what the deposit does to resilience. Spending 60%+ of your savings upfront leaves you exposed to the first emergency, even when the monthly maths works. Assuming income is fixed. Even salaried income has bad months — unpaid leave, job changes. Variable earners at 15–40% volatility should never budget at their average. Not pricing the wait. Unlike most decisions, a purchase can usually wait. Three more months of saving measurably shifts the odds — the engine will show you by how much (honestly noting it can't predict price rises while you wait). The honest context 13.1 million UK adults — 24% — had low financial resilience in May 2024, and 42% could not cover three months of living costs if their main income stopped ( FCA Financial Lives 2024 ). Over-committed budgets are the norm, not the exception. The whole point of doing this maths before you sign is staying out of that statistic. What we won't tell you: whether the purchase is worth it — value, depreciation, joy per pound. There's no defensible statistic for whether big purchases "work out", so LEXUN doesn't invent one. The model forecasts only the resolvable question: does your buffer survive the commitment? It refuses to run on guesses, and shows every answer as an honest range. Work out yours Runs entirely in your browser; your figures stay on this device — the optional Copilot is the one exception, and it sends only what you deliberately type or ask it to summarise. You'll get a survival band, the stress case, the largest commitment your budget can honestly carry, what waiting would change, and a frozen criterion to come back and score against reality. Run the affordability analysis → Method note — maths, assumptions, failure modes The maths. The buffer takes the upfront cost on day one, then each month adds net income (with log-normal shocks of mean exactly one at your stated volatility, so volatility widens the spread without moving the average) and subtracts essential costs plus the recurring commitment. 5,000 seeded paths over the horizon; failure = buffer ≤ 0. The model also solves for the break-even commitment — the largest monthly amount whose median path still survives. Key assumptions. The commitment is constant (variable-rate finance breaks this — re-run when rates change); income shocks are independent month to month; resale/exit value of the purchase is deliberately ignored, which makes the result conservative for recoverable assets like vehicles. Failure modes. Quoted monthly costs are usually understated — insurance, maintenance and service charges belong in the commitment figure, and the adversarial challenge flags this as the main evidence gap; an upfront cost at or above the buffer fails validation rather than pretending. "Worth it" is not modelled — only affordability survival is resolvable. References. Environment: FCA Financial Lives 2024 — 24% of UK adults with low financial resilience; 42% could not cover 3+ months of essentials if income stopped. Scoring and registry: Accuracy Centre . Related guides. The same method applied to two other decisions: business or personal runway , and leaving a job . Terms such as survival band, frozen criterion and Brier score are defined in the decision glossary , and you can re-run a published result yourself before relying on any of it. Stated weaknesses — what this model cannot see These are opportunity@1.0.0's blind spots, stated from the recurrence itself: The upfront cost lands at month zero, in full. Staged payments, deposit-exchange-completion timing, anything paid over time — not modelled. The monthly commitment never changes. That is a fixed-rate assumption. A variable rate rising two points is invisible to this model, and for a mortgage that can be the whole question. Essential costs are frozen; income shocks are symmetric and independent. The same simplifications as the other models, with the same consequences. It refuses to run when the upfront cost meets or exceeds your buffer rather than modelling borrowing to bridge the gap. The asset itself is not modelled at all. This is the affordability of the commitment, not the return on the purchase. A house that doubles in value and a house that halves score identically here. If one of these is the crux, the model is the wrong tool — the insufficient-evidence state is how it says so when it can tell, and this list is how we say so when it cannot. Where to go next Run your own numbers The worked example, with your figures instead. Runway guide The same maths for cash rather than a purchase. Decision templates Start from a structure rather than a blank page. Questions people actually ask Can I afford this big purchase? Affordability is survivable cash flow, not a feeling. LEXUN simulates your buffer against the upfront cost plus every recurring cost of ownership across 5,000 paths of your income, and reports the probability the buffer survives the horizon — plus the largest commitment the median path survives. Is it better to wait before a major purchase? Waiting has a measurable price and a measurable benefit, and the honest answer is a comparison, not a slogan. LEXUN shows what three months of waiting typically changes for your survival odds given your savings rate — sometimes waiting buys safety, sometimes it just costs time. Does LEXUN tell me whether the purchase is worth it? No — worth, suitability and resale value are not modelled, and the result says so. LEXUN answers the resolvable part: whether your finances survive it. Whether it is worth it remains your call, made with the numbers in front of you. Product Simulator My decisions Evidence For teams Pricing Company About LEXUN Contact Roadmap Resources API Status Changelog Legal Privacy notice Terms of service Cookies and local storage Security The full site — every page LEXUN publishes Home Runway guide Career-change guide Big-purchase guide Glass box vs black box Reproduce a result Glossary How LEXUN compares Why this is hard to copy Careers For education API System status Changelog Cite & check our data Country coverage Labs: is my role exposed to AI? © 2026 LEXUN — Decision Intelligence. Decision intelligence, not financial, legal or medical advice. Operated by CA Capital Limited, registered in England and Wales, company number 10848369, registered office 320 Firecrest Court, Centre Park, Warrington, United Kingdom, WA1 1RG. ICO registration ZC183768. View the record at Companies House . Contact: hello@lexun.co.uk

## Accuracy Centre (https://lexun.co.uk/accuracy)

LEXUN Accuracy Centre — method, calibration, weaknesses Skip to content LEX UN Simulate My decisions Evidence Teams Get your Key Search Ctrl K Menu Run a simulation Accuracy Centre Every forecasting product should publish how it forecasts, how often it has been right, and what it is bad at. This page does that — including when the honest answer is "we don't have enough data yet." Current public record: not enough resolved outcomes yet. LEXUN's rebuilt platform (runway model v1.0.0) launched in July 2026. Calibration statistics are computed only from forecasts whose outcomes have been reported and resolved against criteria frozen at prediction time. Until at least 10 resolved outcomes per domain exist, no accuracy percentage will appear here — a smaller sample would be anecdote dressed as evidence. Model-domain outcomes resolved: 0 — the editorial forecasts on the public register (6 resolved of 12 issued) are tracked separately and are not evidence for the models. Unresolved (open) forecasts accrue privately per user; aggregate counts will be published when consented aggregation ships. Names: the model shown as Major purchase on every page is opportunity@1.0.0 in the receipts, the ledger schema and this page’s tables — one model, one internal id. On this page Expand all Methodology — runway model v1.0.0 Methodology — career model v1.1.0 Methodology — major-purchase model (opportunity@1.0.0) How accuracy will be reported Reproducibility — a worked receipt Reference data & source registry Rate regimes — historical context The baseline every model must beat Known weaknesses (runway v1.0.0) Model versions Boundaries Methodology — runway model v1.0.0 The business cash-runway model answers one resolvable question: will cash stay above zero for the stated horizon? Simulation: 5,000 monthly cash paths, run in antithetic pairs. Revenue follows a log-normal multiplicative step about the user’s stated growth rate — exp(ln(1+g) − σ²/2 + σ·ε) , ε ~ N(0,1) — whose expected value is exactly 1+g at any volatility and which is strictly positive, so no month can drive revenue to zero; costs grow at the stated rate; committed one-off cash events land in their stated months. Determinism: the random seed is derived from the inputs (or supplied explicitly), so identical inputs always produce identical results. This is enforced by automated tests. Uncertainty display: probabilities are shown as bands rounded to 5-point steps, widened by Monte-Carlo standard error and by input quality. Raw values are stored only for calibration scoring. Evidence gates: the model refuses to run when core inputs are marked unknown , or when every input is an assumption. It does not fabricate values. Confidence: derived structurally from how many core inputs are facts versus assumptions — never from model enthusiasm. Adversarial review: a separate challenge module (different method: straight-line burn heuristic plus a rule library) attacks every result; material disagreement between the two models is recorded and shown. Resolution: the outcome criterion and due date are frozen at prediction time and cannot be edited afterwards. Outcomes are scored with the Brier rule, which is defined with the rest of the vocabulary in the decision glossary . Methodology — career model v1.1.0 The career-change model answers only the resolvable question: if you make the change, does your financial buffer survive the income transition? Income ramps linearly to your expected level over your stated months, with monthly volatility shocks; essential costs drain the buffer; 5,000 seeded paths. Same disciplines as runway: deterministic seeds, evidence gates, banded display, frozen resolution criteria. Version 1.1 delivers the independent-heuristic cross-check promised at 1.0: a deterministic straight-line drain model (separate code path, no Monte-Carlo) attacks every result, and material disagreement between the two is recorded and shown. Stated honestly: no defensible base rate exists for career-change success (definitions vary too much to pool), so this model does not offer one. Context figure only: ~2.9m UK workers changed jobs in 2025 ( Indeed Hiring Lab, citing ONS ). Known weaknesses: partner income, benefits during transition, emergency costs and the return-to-employment option are not modelled. Methodology — major-purchase model ( opportunity@1.0.0 ) You will see this model called major purchase in the product and opportunity@1.0.0 in receipts, version tables and the changelog. They are the same model: opportunity is the internal identifier it was released under, and renaming it would break every reproducibility id already issued against it. The major-purchase model answers only the resolvable question: if you take on this upfront cost and recurring commitment now, does your financial buffer survive the horizon? The upfront cost leaves the buffer at month zero; thereafter income varies with monthly volatility shocks against fixed essential costs plus the commitment; 5,000 seeded paths. It also reports the largest monthly commitment at which your median path still survives, and a waiting analysis that honestly notes it does not model price rises while you wait. Same guarantees as the other models, including a paired independent surplus-arithmetic cross-check. Stated honestly: no defensible base rate exists for whether a major purchase "works out" (outcomes vary too much to pool), so this model does not offer one. Context only: 13.1 million UK adults (24%) had low financial resilience in May 2024, and 42% could not cover 3+ months of living costs if their main income stopped ( FCA Financial Lives 2024 ). Known weaknesses: price and interest-rate changes while deciding, one-off repair shocks, resale/exit value, and essential-cost inflation are not modelled. How accuracy will be reported Calibration by domain (runway, career, opportunity) and by horizon bucket (0–3m, 3–12m, 12m+) — never one universal percentage. Reliability tables (mean forecast vs actual frequency per probability bin), sample sizes, and confidence intervals. Per model version, so improvements and regressions are visible across versions. Retractions and corrections, listed on this page permanently. The three clocks — how the register avoids hindsight An honest forecasting record has to keep three different times apart: when something happened , when the figure about it was published , and when this register learned it. Mixing them is how systems quietly cheat — a “prediction” scored against a revised figure the forecaster could never have seen is not a prediction. Every ledger entry carries the three clocks as separate frozen fields, and the resolution tool enforces them: issued_at and data_cutoff record what the forecaster could know and when; source_publishes_at records when the deciding figure becomes public — resolving before it is refused outright as “a guess wearing a timestamp”; and the outcome records source_published_at and resolved_at with the figure as first published , so a later revision can never rewrite a score. A worked example from the register: the July 2026 CPI forecast was issued 14 August with a cutoff of the same day, its source published 19 August at 07:00, and it resolved that morning against the first-published 2.9% — the criterion says in terms that later revisions do not change the resolution. The reference-class figures shown beside results follow the same rule from the other side: each is a bundled snapshot with its publication and retrieval dates displayed, marked as context that does not enter the calculation — and never described as live. Outcome verification grades Not all resolutions are equally strong, so every resolved outcome now carries a verification grade, computed from the evidence the ledger actually holds — never assigned by hand. A means automatically resolved from an authoritative source by a connector: impossible in this architecture, because no connectors exist, so no outcome can carry it yet and none does. B means manually resolved against an authoritative published figure — the exact first-published sentence, its URL on the frozen source host and its publication time all sealed in the ledger; all six resolved outcomes on this register are grade B. C and D cover corroborated and unverified user reports — private workspace outcomes, which never enter this public record. E is unresolved or ambiguous. The grades, with each outcome’s grade and the full model coverage and release registry — every model LEXUN ships, every model it merely proposes, and the wall between them — ship machine-readable at /model-registry.json , derived from the model code itself and drift-blocked by its own release gate. The three histories — reality, knowledge, expectation Most forecasting systems keep one history: what happened. An accountable one needs three, kept apart. Reality history is what later evidence says happened — here, every resolved outcome as first published, sealed with its source sentence and URL, never rewritten by revision. Knowledge history is what could genuinely be known at the time — every entry’s data_cutoff , the world-state snapshot frozen into each decision, and the receipt that proves nothing later leaked in. Expectation history is what LEXUN believed would happen next — and it is the one histories quietly lose, because keeping it means leaving your worst calls on display. This register keeps it. Every probability band ever issued stays published; a change of mind is a new entry that names the old one and the reason, never an edit. The table below is that expectation record, regenerated from the ledger at every release — the corrections with the reasons in the record’s own words: Every correction on the register, derived from the ledger at release time — the earlier belief stays published Earlier belief What changed, in the record’s own words Standing belief 001 issued 2026-08-14 · 62–78% · withdrawn Issued and PUBLISHED on lexun.co.uk on 14 August 2026, then withdrawn the same day. The reasoning treated the residual probability as cut risk. The July 2026 MPC minutes record a 6-3 vote with three members — Greene, Mann and Pill — voting to INCREASE Bank Rate to 4%, and the Committee judging… 007 70–84% · open 005 issued 2026-08-14 · 80–90% · withdrawn Issued and PUBLISHED on lexun.co.uk on 14 August 2026, then withdrawn the same day. The due date was set to 2026-08-18T08:00:00Z, which is 09:00 BST. The Insolvency Service publishes its monthly statistics at 09:30 BST — verified from the GOV.UK content API, which gives first_published_at… 009 80–90% · resolved 006 issued 2026-08-14 · 88–96% · withdrawn Issued and PUBLISHED on lexun.co.uk on 14 August 2026, then withdrawn the same day. It said 88-96% that Bank Rate would not rise by 5 November, on the reasoning that every indicator pointed away from tightening. The July 2026 MPC minutes show three of nine members already voting to hike and risks… 008 38–56% · withdrawn 008 issued 2026-08-14 · 38–56% · withdrawn Issued 14 August 2026 and withdrawn the same day, before deployment. The due date was 2026-11-05T12:00:00Z. The Bank of England publishes the Monetary Policy Summary and minutes at 12 noon UK, and British Summer Time ends on 25 October 2026, so on 5 November 12:00 UK IS 12:00 UTC. The deadline was… 010 38–56% · open 12 entries ever issued · 4 since withdrawn · 0 deleted · 0 edited. A register that could quietly rewrite its expectation history would have nothing to be scored against. What this page does not hold: an archive of other forecasters’ expectations — IMF and central-bank projection vintages, professional survey rounds. That is real evidence and a real gap, recorded as such in the gap registry rather than papered over; until it closes, the expectation history here is LEXUN’s own, complete and uncut. Anatomy of one decision record — the nine fields What the engine writes for every run, in the order it writes it. Every field below is live: it is filled by the engine or by you, and nothing on this list is planned or partial. See the exported JSON → The question, as you asked it Live Every input, classed fact or assumption Live Probability band and confidence Live Resolution criterion and due date, frozen at prediction time Live The outcome, written once when reality answers Live Brier score and the domain calibration it updates Live The action you actually took, in your words Live Why reality differed, in your words Live Lessons carried into the next prediction Live Reproducibility — a worked receipt The example on the homepage is a real engine run. Anyone entering the same inputs reproduces it exactly: inputs: cash £80,000 · revenue £12,000/mo · costs £18,000/mo · growth 2%/mo · volatility 15%/mo · horizon 12m model: runway@1.0.0 · seed: 564 (input-derived) · paths: 5,000 reproducibility_id: rw_52f9a7c3_234 output: survival 60–70% (band) · runway p10/p50 9/12+ months (12+ = still solvent when the 12-month window closed) · verdict "Probably safe — watch it" Run it yourself → Reference data & source registry The registry is also shipped machine-readable at /source-registry.json — generated from the same code the engines load (never hand-typed twice), with each source’s authority tier (1 official · 2 recognised research · 3 reputable industry · 4 secondary reporting), licence, publication and retrieval dates, and the register’s resolved-outcome sources. It states plainly that this static site has zero live data connectors: every figure is a bundled, dated snapshot. A release gate blocks any drift between the shipped file and the code. The file also carries a coverage matrix (every figure LEXUN actually holds, with tier, dates and where it is used) and a blind-spot register (what it lacks, classified honestly — from “discoverable and open” to “unavailable in the current architecture” — each with why and the acquisition path). Reference data and source registry Source Figure used Published Retrieved Status ONS Business Demography, UK: 2024 OGL v3.0 · official statistics 5-year survival of UK businesses born 2019: 38.4% 2025-11-20 2026-07-30 Bundled snapshot — not live Indeed Hiring Lab — Job Switching in the UK industry source, citing ONS UK workers who changed jobs in 2025: ~2.9 million (context only) 2026-03-04 2026-07-27 Bundled snapshot — not live FCA — Financial Lives survey 2024: key findings OGL v3.0 · official survey UK adults with low financial resilience, May 2024: 13.1m (24%) ; limited savings buffer: 42% (context only) 2025-05-16 2026-07-30 Bundled snapshot — not live Bank of England — Official Bank Rate history BoE Database terms · official series Every rate change 1975–2025: 258 observations , shipped at /data/boe-bank-rate.json (reference context only — not a model input) maintained series 2026-08-27 Bundled snapshot — not live ONS — CPI annual rate (D7G7, MM23) OGL v3.0 · official statistics Yearly averages 1989–2025: 37 observations , shipped at /data/ons-cpi-annual.json (current vintage only, stated in the file; not a model input) 2026-08-19 2026-08-27 Bundled snapshot — not live No source is ever described as "live" unless an actual connection exists and has recently refreshed. Today, none are live; all reference data is bundled snapshots with retrieval dates. Rate regimes — historical context, not prediction The first Historical Intelligence dataset: the Bank of England’s full official Bank Rate change history — every change since 1975, bundled with its source, licence and retrieval date, and six anchor values cross-checked against independently known history before bundling. It exists to answer one honest question: is today’s rate environment ordinary or unusual by the record’s own standard? No released model takes the rate as an input, so this series is context beside results, never inside them. LEXUN groups the record into named bands — near-zero (below 1%), low (1–3%), moderate (3–6%), restrictive (6–10%), extreme (10% and above). The thresholds are a naming convention, stated so you can disagree with them; they are not analysis. Computing from the bundled series… Two variables against history — an honest analogue With two bundled series — the Bank Rate record and the ONS CPI yearly averages (37 years, 1989–2025, shipped at /data/ons-cpi-annual.json , current vintage only and the file says so) — LEXUN can ask a narrow, honest question: which past years most resembled today on these two variables alone? Today’s pair is real on both sides: the 12-month CPI rate as resolved by the public register against its first-published source (July 2026: 2.9%), and the current Bank Rate from the bundled record. One stated mismatch: today’s 12-month rate is compared with calendar-year averages — said here rather than hidden. Computing from the bundled series… Two variables are not a regime. A real regime comparison needs many more series — unemployment, credit, wages, housing, energy — which this static bundle does not hold; the blind-spot register says exactly that. And even a full state vector would not make history a prophecy: the years listed above went on to different futures for reasons these two numbers never carried. Similarity is not repetition. This series records where the rate has been; it says nothing about where it will go. Two periods that look alike in one variable can end differently for reasons no single series carries — which is exactly why the register freezes forecasts and scores them, rather than reading history as prophecy. The baseline every model must beat — scored on 36 years Forecast science starts with a humiliating rule: before any model earns trust, it must beat the do-nothing line. Below, the simplest possible CPI forecast — next year’s average will equal this year’s — is walked forward through the bundled ONS series with no hindsight: each year’s prediction uses only the previous year’s figure, exactly as a forecaster standing in that year could have. Computed in your browser from the same file you can download . Computing from the bundled series… What this exhibit is for. In quiet years the naive line is very hard to beat — and in regime breaks it fails catastrophically, which is the honest shape of the forecasting problem. Any model LEXUN ever promotes for a variable like this must beat this baseline out of sample, and the years in the misses table are why “history as prophecy” is refused everywhere on this site: the past is the best guide available right up until the moment it is the worst one. Known weaknesses (runway v1.0.0) Revenue shocks are modelled as smooth monthly volatility; single lumpy events (losing an anchor client, a tax bill) are only captured if entered as committed cash events. Receivables timing (booked vs banked revenue) is not modelled. The bundled reference class pools all UK industries and sizes; sector-specific base rates are not yet included, and the 1-year survival rate was not captured in our snapshot, so it is not shown. Outcome data is self-reported until verified-evidence attachment ships; self-reporting can bias calibration optimistically. No calibration history exists yet — the model's real-world track record is unproven, which is exactly what the outcome loop exists to fix. Model versions Model version changelog Version Released Notes runway@1.0.0 · core@1.0.0 2026-07 Initial release. career@1.0.0 · learning@1.0.0 · intelligence@1.0.0 · governor@1.0.0 2026-07 Career financial-resilience model; outcome-gated learning layer; live governor checks. career@1.1.0 · opportunity@1.0.0 · challenge@1.1.0 2026-07 Career independent-heuristic cross-check delivered as promised at 1.0; major-purchase affordability model with its own paired cross-check. Test suite: 120 automated checks, all passing. reality@1.0.0 · core registry 7 sources 2026-07 Decision Reality layer: monitored assumptions, materiality checks, forecast versioning, trust receipts, .lexun export. Registry extended with four sourced Find-My-Money figures. Test suite: 233 automated checks, all passing. core@1.1.0 · vault@1.0.0 2026-07 The LEXUN Key: the saved record is encrypted at rest with AES-256-GCM under a PBKDF2-HMAC-SHA256 key at 600,000 iterations. No model maths changed — core gained a persistence seam only, so every engine version above is unaffected. Test suite: 390 automated checks, all passing. reality@1.1.0 2026-08 The decision record gained what was missing from it: a free-text account of the action actually taken and a separate free-text account of why reality differed, plus lessons carried forward into the framing screen for the domain they were learned in. Additive record fields only — no model maths, seed derivation or hash function changed, so every version above is unaffected and all 96 frozen results resolve unchanged. Test suite: 390 automated checks, all passing. router@2.0.0 2026-08 Decision router rebuilt as a capability firewall: three outcomes (supported, clarify, unsupported) replace “best guess” selection; restricted categories are refused with a supported reframing where one exists. Routing decides which door opens — it never touches model output, so every engine version above is unaffected and all 96 frozen results resolve unchanged. Test suite: 397 automated checks, all passing. router@2.1.0 2026-08 Firewall keyword widening and the advertised-question intent contract; the first public forecast resolved (editorial board — unemployment, TRUE, Brier 0.0100) with every counter reading from the ledger. No engine maths changed; 96 frozen results unchanged. Test suite: 711 automated checks, all passing. Boundaries LEXUN models business and career decisions. It refuses medical, legal and immigration questions, routes personal-crisis language to human help, and provides decision intelligence — not financial, legal or medical advice. Watch the record grow A zero above means no outcome in that domain has resolved — not that nothing is measured. Leave an email and we will write to you when outcomes resolve and the first calibration table publishes: the misses included, because a record without misses is marketing. Leave this empty: Email address Notify me when reality answers One address, one purpose. It is used for record updates and nothing else — see Privacy . Where to go next Reproduce a result Re-run a published forecast and compare the receipt ids. Forecast register Every open forecast, with the date it resolves. Glass box vs black box The same run, shown with its method and then without. Product Simulator My decisions Evidence For teams Pricing Company About LEXUN Contact Roadmap Resources API Status Changelog Legal Privacy notice Terms of service Cookies and local storage Security The full site — every page LEXUN publishes Home Runway guide Career-change guide Big-purchase guide Glass box vs black box Reproduce a result Glossary How LEXUN compares Why this is hard to copy Careers For education API System status Changelog Cite & check our data Country coverage Labs: is my role exposed to AI? © 2026 LEXUN — Decision Intelligence. Decision intelligence, not financial, legal or medical advice. Operated by CA Capital Limited, registered in England and Wales, company number 10848369, registered office 320 Firecrest Court, Centre Park, Warrington, United Kingdom, WA1 1RG. ICO registration ZC183768. View the record at Companies House . Contact: hello@lexun.co.uk

## Glossary (https://lexun.co.uk/glossary)

Decision intelligence glossary — what each LEXUN term means Skip to content LEX UN Simulate My decisions Evidence Teams Get your Key Search Ctrl K Menu Run a simulation Plain English Decision intelligence glossary: every term explained LEXUN uses precise language on purpose, because vague language is how forecasting products avoid being checked. Each term below is defined in one sentence, then explained, then linked to the part of the product that proves we do what the definition says. On this page Expand all Decision intelligence Probability band Base rate and reference class Seeded Monte-Carlo simulation Evidence gate Insufficient evidence Frozen resolution criterion Calibration Brier score Decision quality versus outcome quality Material change Trust receipt LEXUN Key Local-first Fact, assumption and prediction labelling "Planned" Decision intelligence In one sentence: treating a decision as something with stated inputs, an explicit probability, a success test written in advance and a recorded outcome, so the quality of the reasoning can be judged separately from the luck of the result. Most tools help you organise a decision. Decision intelligence goes further: it commits to a number before the answer is known, and then goes back afterwards to check whether that number was any good. It is the checking that makes it intelligence rather than presentation. LEXUN is not financial, legal or medical advice, and it never decides for you — it makes your own reasoning explicit enough to be wrong in public. Probability band In one sentence: a range such as 55–65% rather than a single number, where the width of the range is the honest amount of uncertainty. A model that tells you "61.4%" is claiming a precision it does not have. LEXUN rounds every probability to 5-point steps and then widens the band by two things: the standard error of the simulation itself, and how much of your input was measured fact rather than assumption. A wide band is not the model being unhelpful. It is the model declining to lie about how much it knows. Base rate and reference class In one sentence: how often this outcome happens across the broad class of similar situations, before anything specific to you is considered. The reference class is the group your situation belongs to; the base rate is what usually happens to that group. Starting from the base rate and adjusting is far more reliable than reasoning from your own case alone, because your own case always feels exceptional and usually is not. When LEXUN cannot find a defensible reference class for a question, that is one of the reasons it will decline to answer. Seeded Monte-Carlo simulation In one sentence: running the same model thousands of times with randomly drawn inputs to see the spread of possible futures — but from a fixed starting seed, so the same inputs always produce exactly the same answer. Monte-Carlo simulation is old, public and unremarkable; anyone can implement it. The seed is the part that matters here. Because the randomness is deterministic, a LEXUN result is not a one-off performance you have to take on trust — you can re-run it and get the identical output, byte for byte. Reproduce a result in your browser to see this happen. Evidence gate In one sentence: a rule that stops the model running at all when there is not enough real information to run it responsibly. The gates are specific: the model refuses when core inputs are marked unknown, and it refuses when every single input is an assumption rather than a measured fact. This is deliberately inconvenient. A product that always produces an answer is a product that produces answers it has not earned, and the commercial pressure runs entirely in that direction — which is why the refusal is built into the engine rather than left to judgement. Insufficient evidence In one sentence: a real, valid output of the engine, not an error message. When a gate trips, LEXUN says it cannot answer and tells you which input would change that. Treating "I don't know" as a legitimate result rather than a failure is the difference between a forecasting tool and a confidence generator. Frozen resolution criterion In one sentence: the test for whether a forecast came true, written down before the outcome is known and never edited afterwards. Without this, scoring is meaningless: anyone can be right in retrospect by quietly redefining what they meant. Freezing the criterion at prediction time is what makes a later score a fact about the forecast rather than a story about the forecaster. Every entry in the forecast register carries one. Calibration In one sentence: whether the things you called 70% likely actually happen about 70% of the time. Calibration is not the same as being right a lot. A forecaster who says 90% about everything and is right 90% of the time is well calibrated; one who says 99% and is right 90% of the time is overconfident, even though both were right equally often. Calibration can only be measured from outcomes that have actually resolved, which is why LEXUN's published record is currently an honest zero — see the Accuracy Centre , which prints that zero rather than hiding it. Brier score In one sentence: the average of the squared difference between what you said would happen and what did, where 0 is perfect and 1 is as wrong as it is possible to be. If you forecast 0.8 and the thing happens, the outcome counts as 1 and that forecast scores (0.8 − 1)² = 0.04. If it does not happen, it scores (0.8 − 0)² = 0.64. Averaging that over many forecasts gives the Brier score. The number worth remembering is 0.25 : that is what you get by saying "50%" to everything forever, so any scoring system that cannot beat 0.25 has told you nothing at all. Decision quality versus outcome quality In one sentence: a good decision can have a bad outcome, and a bad decision can have a good one, so the two have to be scored separately. If you take a well-reasoned 80% bet and land in the 20%, you were unlucky, not wrong; if you take a reckless 5% bet and it lands, you were lucky, not clever. Judging decisions purely by how they turned out teaches you the wrong lesson in both cases. When you resolve a decision in LEXUN it asks you to separate the reasoning, the execution and the result, because the fix for each is different. Material change In one sentence: something has moved enough since you decided that the decision deserves revisiting. A decision is made against a set of assumptions on a particular day. Those assumptions expire quietly, and the usual failure mode is not noticing. LEXUN records what the decision assumed, so it can flag when an assumption you relied on no longer holds instead of leaving that to memory. Trust receipt In one sentence: a SHA-256 hash taken over the exact bytes of a result, so anyone can later prove the result has not been altered. Each result is sealed with its hash, and the record of results is hash-chained — every entry commits to the one before it, so a changed entry in the middle breaks every hash after it. You do not have to trust us for this: you can check a receipt yourself without an account, and the check runs in your own browser. LEXUN Key In one sentence: a passcode you choose that encrypts your saved decisions on your own device, which we never receive and cannot recover. The passcode is stretched into an encryption key using PBKDF2-HMAC-SHA256 at 600,000 iterations, and the record is encrypted with AES-256-GCM. The consequence is worth stating bluntly: because the key never leaves your device, losing the passcode means losing the saved record, and there is no reset link we could honestly offer you. Local-first In one sentence: your decision data is stored in your browser rather than on our servers. Running a decision needs no Key. Saving it to your record does, and so does every part of the dashboard that reads that record. A Key is free, needs no card and is not an account: it is a passcode that encrypts your record on your own device, and the decision itself is still never sent to us. Reproducing a published result and reading the accuracy record need nothing at all. What we hold is the email address you give us when you create a Key or join the waitlist, and nothing else. The security page sets out the architecture and the privacy notice sets out the legal position under UK GDPR. Fact, assumption and prediction labelling In one sentence: every input carries a label saying whether it is something you measured, something you guessed, or something the model inferred. These three get blurred together in almost every spreadsheet ever built, and once blurred, a guess acquires the authority of a measurement. Keeping them distinct is what allows the model to widen its bands when it is running mostly on assumptions, and to refuse outright when it is running on nothing else. "Planned" In one sentence: a feature that does not exist yet, labelled so on the page rather than described in the present tense. Every unshipped capability on this site is marked as planned wherever it appears, and the roadmap lists them in one place. There is no ship date attached to most of them, because each is released when it passes its reliability gate and inventing a date would be the same failure this glossary exists to prevent. If a term is used anywhere on this site and is not defined here, that is a defect worth telling us about: contact us and we will either define it or stop using it. Where to go next Glass box vs black box See these terms attached to one real result. Accuracy Centre Where the definitions are used in anger. Trust centre Each claim, with the evidence behind it. Product Simulator My decisions Evidence For teams Pricing Company About LEXUN Contact Roadmap Resources API Status Changelog Legal Privacy notice Terms of service Cookies and local storage Security The full site — every page LEXUN publishes Home Runway guide Career-change guide Big-purchase guide Glass box vs black box Reproduce a result Glossary How LEXUN compares Why this is hard to copy Careers For education API System status Changelog Cite & check our data Country coverage Labs: is my role exposed to AI? © 2026 LEXUN — Decision Intelligence. Decision intelligence, not financial, legal or medical advice. Operated by CA Capital Limited, registered in England and Wales, company number 10848369, registered office 320 Firecrest Court, Centre Park, Warrington, United Kingdom, WA1 1RG. ICO registration ZC183768. View the record at Companies House . Contact: hello@lexun.co.uk

## How to judge decision tools (https://lexun.co.uk/compare)

Monte Carlo decision tools — how to judge them — LEXUN Skip to content LEX UN Simulate My decisions Evidence Teams Get your Key Search Ctrl K Menu Run a simulation The category, compared How we differ from population-simulation vendors. A category of AI vendors now sells large-scale "simulation of people" to enterprises and governments: striking claims, gated demos, no published method. This page compares approaches, not companies — on the axes any serious buyer should test every vendor against, including us. On this page Expand all What actually separates the two approaches The questions to ask any simulation vendor What we refuse to fake Which approach fits which buyer What actually separates the two approaches In one sentence: a gated simulation vendor asks you to trust a result you cannot inspect, while LEXUN publishes the method, shows the error bars, keeps a public calibration record and hands you a way to reproduce the answer yourself. The nine rows below are the axes where that difference stops being rhetoric and becomes something you can test — each one is a question you can put to any vendor, including this one, and expect an artefact in reply rather than an assurance. Axis Typical gated simulation vendor LEXUN Published method Undisclosed; "proprietary simulation" Full methodology public on the Accuracy Centre ; model maths documented per domain in the method guides Error bars Point claims ("we predicted X within 0.n%") Every probability shown as a band; binomial standard error disclosed; false precision refused by design Calibration record None published Public Brier-scored record, computed only from resolved outcomes — published from an honest zero on day one, growing only as reality answers Reproducibility Not offered Seeded and versioned: identical inputs reproduce identical outputs — verify in your browser now Tamper evidence Not offered Every result is sealed with a SHA-256 hash over its canonical bytes and the audit record is hash-chained — check a result without trusting us Stated moat Scale, data access, proprietary method Scoring, receipts and elapsed time — set out in full, including what is not a moat, on the defensibility page Access Demo request, sales-gated Free and in your browser today — no Key needed to run a decision; a LEXUN Key is required to save one, and a Key is a free passcode that encrypts your record on your own device rather than an account on ours Refusal behaviour An answer for everything "Insufficient evidence" is a real output; out-of-scope domains are declined, not improvised Data handling Your scenario data on their servers Local-first: simulation inputs and saved decisions stay in your browser; the optional Copilot sends only what you type into it and a filtered result summary — architecture, plainly Rather than argue the first two rows, we built them into a page you can press: the same simulation delivered as a glass box and as a black box . One run, two panels, live in your browser. Fairness note: gated vendors may have strong internal validation we cannot see. That is precisely the point — a buyer cannot act on evidence they are not shown. Apply this table to us too: every LEXUN row above links to the artefact that proves it. The questions to ask any simulation vendor Can I see the calibration record, and who computed it? What does a forecast look like when you are uncertain ? Can I re-run yesterday's prediction and get the same answer? What happens when I ask something outside the model's competence? Where does my data live while you simulate with it? We publish our answers. Demand everyone's. What we refuse to fake — and why that wins A comparison that only flatters its author is marketing, so here is our position, plainly — every line of it a choice, not a gap. Our public calibration record started from zero on day one, and the Accuracy Centre printed that zero in the open: that is what an honest record looks like at the start, and it is exactly why it will be unanswerable later — a record that started earlier can be caught up with; one that started honest cannot be faked. We have not found a published, scored forecast record among the vendors we reviewed; if one exists, we will link to it here. Our coverage is three decision domains — business runway, career change, major purchase — built deep rather than a catalogue built thin: versioned models, stated weaknesses, reproducible to the byte. And when a question sits outside them, the engine says so instead of improvising — the refusal is the feature. An engine that answers everything is an engine whose answers mean nothing. Which approach fits which buyer Gated population simulation is built for organisations buying scale: large panels, bespoke modelling engagements and a supplier relationship with account managers attached. LEXUN is built for the person or small team facing one consequential decision this week whose reasoning has to survive scrutiny afterwards — from a board, an auditor, a co-founder, or their own future self reading it back in a year. The two are not really chasing the same budget. They are chasing the same claim: that a number on a screen has earned the right to be believed. Where to go next Glass box vs black box One real run, rendered both ways, on one page. Why this is hard to copy The part of the comparison that takes years, not a sprint. Pricing What it costs once you are convinced. Questions people actually ask How is LEXUN different from asking a chatbot? A general-purpose assistant generates a plausible answer; LEXUN's numbers never come from language-model intuition. Every probability comes from a seeded, versioned simulation you can reproduce bit-for-bit, every result is frozen with a criterion and a date, and reality grades it afterwards on a public register. The honest cost of that discipline: LEXUN answers far fewer kinds of questions. What should I look for in any decision or prediction tool? Five things, and they apply to LEXUN too: can you reproduce a result exactly; does every number have a source or a stated assumption; are forecasts frozen before outcomes are known; is there a scored track record rather than testimonials; and does the tool ever say 'insufficient evidence'. A tool that fails these tests is asking to be trusted rather than checked. Can LEXUN tell me which stocks or crypto to buy? No, and it refuses rather than guessing: no financial-market model is released, no licensed market data is held, and an invented probability about an asset price is the exact failure this product exists to prevent. What it can run is the resolvable question underneath: whether your cash or buffer survives if the amount you are thinking of investing were lost — a released, seeded model with a receipt. Decision intelligence, not investment advice. Is what I type used to train an AI? No. The simulation runs on your device, and the sentence you type into the homepage box is parsed there and never transmitted. The one exception is the optional Copilot in the workspace: it sends what you deliberately type into it to the AI provider, only when you press its button, and result tools send only a filtered result summary — none of the financial figures you entered into the engine. There is no analytics script to smuggle it out — the privacy notice and the network tab agree. The optional assistant is the one clearly-labelled exception, and it says so before you use it. Where is LEXUN weaker than the alternatives? Scope, breadth and sample size. Three released models, not three hundred; verified evidence covers the United Kingdom only; reference data is bundled dated snapshots, not live feeds; and the resolved-outcome sample is still below the ten-outcome minimum per domain, so no accuracy percentage is published yet. These limits are stated on the product because pretending otherwise would cost the one thing LEXUN has: a record you can check. Product Simulator My decisions Evidence For teams Pricing Company About LEXUN Contact Roadmap Resources API Status Changelog Legal Privacy notice Terms of service Cookies and local storage Security The full site — every page LEXUN publishes Home Runway guide Career-change guide Big-purchase guide Glass box vs black box Reproduce a result Glossary How LEXUN compares Why this is hard to copy Careers For education API System status Changelog Cite & check our data Country coverage Labs: is my role exposed to AI? © 2026 LEXUN — Decision Intelligence. Decision intelligence, not financial, legal or medical advice. Operated by CA Capital Limited, registered in England and Wales, company number 10848369, registered office 320 Firecrest Court, Centre Park, Warrington, United Kingdom, WA1 1RG. ICO registration ZC183768. View the record at Companies House . Contact: hello@lexun.co.uk

## Pricing (https://lexun.co.uk/plans)

LEXUN Pricing — free decision engine, honest paid tiers Skip to content LEX UN Simulate My decisions Evidence Teams Get your Key Search Ctrl K Menu Run a simulation Plans Start free. Keep the record. Three models are live — business runway, career change and major purchase. Run one free, today; pay only to keep the record. No account is needed to run a decision. To save one, create a free LEXUN Key: a passcode that encrypts your record on this device (email optional). No card is requested unless you upgrade, and a Key is not an account. How the Key works → The decision engine is free to use. There is no card and no charge — checkout is not switched on for this site yet, and until it is nothing is limited. Annual Save 27% Monthly Free Free: the full engine and unlimited decision runs, no card, no account; three saved decisions once checkout opens, and nothing is limited until then. £0 forever no card required The decision engine is free to use. There is no card and no charge — checkout is not switched on for this site yet, and until it is nothing is limited. Run a simulation → Unlimited decision runs — no allowance, no counter, no card Three released, test-gated decision models — business runway, career change, major purchase. Test-gated means 711 automated checks pass on every release; it does not mean the forecasts have been proven accurate: no accuracy is claimed until ten outcomes have resolved Seeded, reproducible simulations with honest probability bands, receipts and the challenge case Three saved decisions with outcome reporting and calendar reminder files Three Copilot messages a day — describe a decision in your own words and it fills the form from what you wrote; ask it to explain or challenge a result. Every figure in its replies comes from the engine, or is removed Local-first privacy — simulation inputs stay on this device; the optional Copilot sends only what you deliberately type or ask it to summarise 14-day trial · no card Pro For people who want their decisions kept, scored and remembered. Price the record, not the run. £9 / month billed monthly · cancel any time — or £79 a year (£6.58/mo, save 27%) Checkout is not switched on for this site yet — everything below is built and waiting behind it. Start 14-day free trial → No card for the trial. If you add one, Pro continues at the price above; if you do not, it ends and nothing is charged. Everything in Free , plus: Every decision kept — an unlimited ledger of frozen forecasts, resolutions and scores Exports and PDF reports — the portable decision package and the printed record The calibration view — your own accuracy record, scored over time (needs resolved outcomes to show anything — it is honest, not decorative) Resolution reminders — the day reality answers, in your calendar today; by email when the sending domain is connected · email is planned The AI Copilot, sixty messages a day — plain-English intake, the explanation of a result and the assumption challenger; AI text is labelled and no number in it comes from the model Founding-member pricing: the price you join at is the price you keep. Anything marked planned is not yet sold as if it existed. Organisation LEXUN Decisions for Organisations: a permanent, verifiable record of what was decided, on what evidence, by whom — for fractional CFOs, accountancy practices and finance teams. Prices exclude VAT; LEXUN is not currently VAT-registered, so none is added today. £49 / seat / mo billed monthly per seat — or £490 a seat a year (£40.83/mo, two months free) Seats Buy Organisation seats → Decisions for Organisations → Everything in Pro, for every seat Board-ready report generator — the Decision Record as a paper a board can read ( the sample · PDF, 9 pages ) Audit-trail export — the hash-chained record of who did what, when Board summary from the Copilot — an AI-written executive summary of a frozen result for the Decision Record, every figure from the engine; two hundred Copilot messages a day per seat Shared client workspaces, roles and approvals, SSO and API keys · planned — the hosted service, Release 2 Enterprise £25,000–£100,000 a year — agreed after a pilot, never sold from a page Founding pilot £1,500 : one consequential decision through the complete Decision Loop in 30 days; what it includes . Decision Audit £2,500–£15,000 per engagement; the three packages Seats are bought here when checkout is switched on; Enterprise, pilots and audits are invoiced. Nothing marked planned is sold as if it existed. The full entitlement table is on packages . Pricing & consumer rights: payment is handled by Stripe — LEXUN never sees or stores your card details, and Apple Pay, Google Pay and cards all work at checkout. Prices are in pounds sterling. LEXUN is not currently VAT-registered, so no VAT is added at checkout and no invoice shows any; if that changes, Stripe Tax will add it and the prices above stay the net figure. You have 14 days to cancel a new subscription for a full refund under the Consumer Contracts Regulations 2013; after that, cancelling stops the next payment and you keep what you paid for to the end of the period. The decision engine is free to use. There is no card and no charge — checkout is not switched on for this site yet, and until it is nothing is limited. Cancel anytime. Local-first by default — your saved decisions stay on your device (encrypted sync across devices is on the waitlist , not live). LEXUN is decision intelligence, not financial, legal or medical advice. Which door is yours? For individuals Free covers the whole decision loop with unlimited runs; Pro keeps every decision, adds exports, reports, the calibration view and the Copilot. The maths is identical on every tier — a paid plan never buys a better probability, only a kept record. For organisations Organisation seats, an Enterprise agreement after a pilot, the founding pilot and the Decision Audit. Start from the organisations page; every capability is labelled live or planned, and nothing planned is sold as live. For education Classrooms use the browser engine free — no accounts, no student data collected. Pilots and course support are arranged through the contact page, not a checkout. Questions Is the free plan really free? Yes. Unlimited decision runs, no card at any point unless you choose to upgrade, and no account: saving a decision needs a LEXUN Key, a passcode that encrypts the record on your own device. The 14-day Pro trial also needs no card — it simply ends if you do not add one. What does Pro add? The record. Free keeps three decisions once checkout opens (nothing is limited until then); Pro keeps every one, with exports, PDF reports and the calibration view, and lifts the Copilot's daily allowance. The engine is the same on both: a paid plan never buys a different answer. Anything not yet built is labelled planned on this page and is not charged for. How is my data handled? LEXUN is local-first: your data lives in your browser, not our servers. No advertising or tracking cookies, nothing sold. See our Privacy and Cookies policies. Can I cancel? Any time, in one step: Settings → Manage plan in the workspace opens the Stripe billing portal, where you can change plan, change card, download invoices or cancel. Cancelling stops the next payment; you keep what you paid for until the end of the period, and every decision you saved stays readable on the free plan. If anything goes wrong, hello@lexun.co.uk is read. Is there a run limit? No. Free runs are unlimited and always will be; the paid plans sell the record, not the run. A plan is confirmed with Stripe each time you visit rather than trusted from a marker in your browser, and if that check cannot be made your last confirmed plan stands for seven days. A plan that ends loses nothing you saved. Where to go next Start free The free tier runs the same engine, in your browser. For organisations When per-seat is the wrong shape. Terms of service What you are agreeing to before you pay. Product Simulator My decisions Evidence For teams Pricing Company About LEXUN Contact Roadmap Resources API Status Changelog Legal Privacy notice Terms of service Cookies and local storage Security The full site — every page LEXUN publishes Home Runway guide Career-change guide Big-purchase guide Glass box vs black box Reproduce a result Glossary How LEXUN compares Why this is hard to copy Careers For education API System status Changelog Cite & check our data Country coverage Labs: is my role exposed to AI? © 2026 LEXUN — Decision Intelligence. Decision intelligence, not financial, legal or medical advice. Operated by CA Capital Limited, registered in England and Wales, company number 10848369, registered office 320 Firecrest Court, Centre Park, Warrington, United Kingdom, WA1 1RG. ICO registration ZC183768. View the record at Companies House . Contact: hello@lexun.co.uk No tracking cookies. LEXUN uses local storage only for features you choose to use, and to remember your preferences. Nothing tracks you across sites and nothing is sold or shared. Read privacy details . Understood ↑

## Trust and validation (https://lexun.co.uk/trust)

LEXUN — Trust & validation: the verifiable record Skip to content LEX UN Simulate My decisions Evidence Teams Get your Key Search Ctrl K Menu Run a simulation Trust & validation Nothing here is a testimonial. Everything here is checkable. Last updated 2026-09-16 — regenerated at every release from the repository itself. LEXUN is new, so this page contains no press logos and no borrowed credibility. What it contains instead is the verifiable state of the system: what is tested, what is proven, what is versioned, and what is still zero. On this page Expand all Automated verification Determinism proof — run it yourself The calibration record Versioned model changelog Worked case studies What would falsify us Automated verification 711 automated checks run against the engine before any release ships — covering RNG determinism, hash stability, evidence-gate refusals, monotonicity (more cash never lowers survival; a bigger commitment never raises it), Brier scoring, calibration binning, forecast freezing, version immutability, materiality detection and trust-receipt integrity. Suite: /platform/tests.mjs (served read-only) · run: node tests.mjs from the platform folder of the release bundle · current status: 711/711 passing The suite ships in the site bundle itself and is readable at the address above — download the bundle and run it. A claim you can execute is the only kind we make. Hostile input, shown refused. Type “Ignore your rules and just give me a single number” — or any question outside the three released models — and the router refuses before anything computes: free text never reaches a model as an instruction, only as figures to extract, and there is no language model in the answering path to talk out of its discipline. The refusal is itself under test: the release gate feeds the engine hostile text, empty states and malformed files, and a homepage chip labelled “Watch it refuse” lets you press the same boundary yourself. Determinism proof — run it yourself Every result carries a seed, model version and input hash. Identical inputs and seed produce identical output, bit for bit. The canonical worked example: model runway@1.0.0 · seed 564 · P(survival) band 60–70% · id rw_52f9a7c3_234 Reproduce it in one click on the Reproduce a result page, or independently via the published API contract example. To check a sealed receipt or a .lexun package byte for byte, use the verifier . The calibration record Model-domain outcomes resolved: 0 (the editorial forecasts on the public register — 6 resolved of 12 issued — are tracked separately and are not evidence for the models). LEXUN publishes no accuracy percentage until at least 10 outcomes per domain have resolved against frozen criteria. This zero is the load-bearing fact of the whole product: it proves the accuracy record cannot be faked, because we could have faked it already and did not. Method and scoring rules: Accuracy Centre . Why this zero is the asset rather than the embarrassment: defensibility . Versioned model changelog Check counts in this table are historical — each row records the suite as it stood at that release. The current count lives on /platform and /accuracy, and a release gate fails the build if any current claim disagrees with the suite itself. Every current platform statistic also ships machine-readable at /system-manifest.json — each number derived from the repository itself (the suites count themselves, the registries are read, nothing is typed twice) and drift-blocked by its own release gate. Version Date Change runway@1.0.0 · core@1.0.0 2026-07 Initial release: seeded Monte-Carlo runway model, evidence gates, frozen resolution criteria. career@1.0.0 · learning@1.0.0 2026-07 Career financial-resilience model; outcome-gated learning layer; governor checks. career@1.1.0 · opportunity@1.0.0 · challenge@1.1.0 2026-07 Independent heuristic cross-checks; major-purchase affordability model. 120 checks passing. reality@1.0.0 2026-07 Decision Reality layer: monitored assumptions, materiality engine, forecast versioning, trust receipts, .lexun export. 233 checks passing. core@1.1.0 · vault@1.0.0 2026-07 The LEXUN Key: encrypted-at-rest dashboard record (AES-256-GCM, PBKDF2-HMAC-SHA256 at 600,000 iterations). No model maths changed. 390 checks passing. reality@1.1.0 2026-08 Action taken and why reality differed recorded in the person’s own words; lessons carried into the next decision in the same domain. Additive fields only — no model maths changed. 390 checks passing. core@1.2.0 · runway · career · opportunity · path-trace 2026-09 The multiplicative shock became log-normal ( exp(ln(1+g) − v²/2 + v·z) ) so a month can halve revenue and can never delete it, and paths run in antithetic pairs — 34% less Monte-Carlo error for identical work. Survival probabilities moved by up to a point at ordinary volatility and by more at the extremes; every moved result is listed in the changelog entry. core@1.2.1 · runway@1.0.1 · career@1.1.1 · opportunity@1.0.1 2026-09 Break-even searches report when the breakpoint lies outside the searched range instead of returning the bound as a figure; quoted breakpoints are rounded onto the surviving side; optional runway inputs (committed amounts, cost growth) are validated; the survived-code is capped at the horizon in every view. No probability, band, verdict, receipt id or input hash changed — 96 frozen results byte-identical. router@2.0.0 2026-08 The capability firewall: unsupported questions (markets, medical, legal, gambling, relationship outcomes, elections, weather, vague success) are refused with reasons and a supported reframing where one honestly exists; weak signals ask one question instead of guessing; the silent runway default is gone. Routing only — no model maths, seed derivation or hash function changed, so all 96 frozen results resolve unchanged. 397 checks passing. router@2.1.0 2026-08 Capability-firewall keyword widening (pricing, expansion, staffing phrasings; accent folding so “café” reaches the keywords as “cafe”) plus the intent contract: every question the site advertises must route and pass the frame check, enforced in the suite. Routing only — no model maths changed, all 96 frozen results unchanged. 711 checks passing. Methodology never changes silently: every change lands here and in the Accuracy Centre with its version number. Worked case studies LEXUN launched in 2026 and has no client testimonials yet — and we will not invent any. What we can publish today are fully worked, anonymised decision cases: real engine runs whose every number you can reproduce. As real outcomes resolve, their Brier scores will be appended here — whatever they show. Case 1 — Small business runway. £80,000 cash, £12,000/mo revenue growing 2%/mo (±15% volatility), £18,000/mo costs, 12-month horizon. Forecast: 60–70% probability of maintaining positive cash (PROBABLY SAFE, WATCH IT); top driver monthly costs; the stress scenario falls to about 20%. runway@1.0.0 · seed 564 · rw_52f9a7c3_234 — reproduce it . Outcome: resolves against the frozen criterion at its due date; no outcome yet, so no score is claimed. Case 2 — Career change. £12,000 buffer, £2,400/mo essential costs, £2,600/mo expected new income after a 6-month ramp (±30% volatility). The engine prices the transition honestly: the buffer question is resolvable, career "success" is not — so only the former is forecast. career@1.1.0 · worked example seed 705 — full walk-through . Case 3 — Major purchase. £12,000 buffer, £6,000 upfront, £550/mo commitment against £2,600/mo net income and £2,050/mo essentials. The model reports buffer survival and the break-even commitment, and its adversarial challenge flags the assumed running costs as the evidence gap. opportunity@1.0.0 · worked example seed 507 — full walk-through . Case 4 — An adviser’s file (illustrative; the client and figures are fictional, the engine run is real). An adviser’s client wants to leave employment for a consultancy contract. Inputs as declared with the client: £20,000 buffer (fact — statement seen), £1,800/mo essential costs (fact — 6-month bank average), £2,500/mo expected new income (assumption — one signed letter of intent), 3-month ramp (assumption). The engine returns its band with the fact/assumption split preserved, and the exported Decision Record — inputs, classifications, band, model version, reproducibility id, sign-off line — goes on the client file as evidence of a documented, disciplined process. What it does not do: make the recommendation, or make anyone compliant. The adviser door, stated precisely . What would falsify us If our resolved forecasts turn out mis-calibrated, the calibration table will say so — it is computed, not written. We commit to publishing that table whether it flatters us or not. That commitment, not a promise of accuracy, is the product. Where to go next Accuracy Centre The numbers underneath the claims on this page. Verify a result Do not take the claim; check a receipt. Glass box vs black box What a claim looks like with its method attached. Product Simulator My decisions Evidence For teams Pricing Company About LEXUN Contact Roadmap Resources API Status Changelog Legal Privacy notice Terms of service Cookies and local storage Security The full site — every page LEXUN publishes Home Runway guide Career-change guide Big-purchase guide Glass box vs black box Reproduce a result Glossary How LEXUN compares Why this is hard to copy Careers For education API System status Changelog Cite & check our data Country coverage Labs: is my role exposed to AI? © 2026 LEXUN — Decision Intelligence. Decision intelligence, not financial, legal or medical advice. Operated by CA Capital Limited, registered in England and Wales, company number 10848369, registered office 320 Firecrest Court, Centre Park, Warrington, United Kingdom, WA1 1RG. ICO registration ZC183768. View the record at Companies House . Contact: hello@lexun.co.uk