About When Can I Retire?
Working out when you can afford to retire means pulling together pensions, savings, investments, property, tax and decades of compounding — the kind of sum most people never get a clear answer to. When Can I Retire? does that maths for you: enter your details, or upload a spreadsheet, and see your earliest retirement date, the income your plan sustains, and how it holds up when markets, inflation or longevity don't go to plan.
How it works
A projection engine simulates your household month by month from today to your plan age: income from work, pensions and rentals, minus tax and spending, with surpluses invested and shortfalls drawn from your pots in the order you choose. It has UK tax built in — income tax bands, the 25% tax-free lump sum, National Insurance, capital gains and an inheritance tax estimate, on verified 2026/27 figures. Goal solvers find your earliest possible retirement date and the income your plan supports; Monte Carlo and sensitivity tests show how robust it is. The engine runs entirely in your browser — your figures never leave your device unless you sign in to save them or use the optional AI features. Every assumption and source is on the methodology page, and an audit view inside the planner shows exactly how each headline figure was calculated.
How this site was built
There is no development team behind When Can I Retire?. It was designed, built and is run by one person directing AI tools — and this page exists to show what that means in practice.
The division of labour:
Claude (claude.ai) acts as architect and operations engineer. Through live connections to the infrastructure (the Model Context Protocol, MCP), it designs the systems and then runs them in conversation: provisioning the database, applying schema changes, reading deployment logs, diagnosing build failures and fixing them.
Claude Code writes the site itself: the projection engine, every page and component, the AI features and the tests — working from written briefs, verifying its own work against live builds, and pushing to production.
Supabase handles accounts: Google sign-in, a per-user table that syncs your saved plans across devices with row-level security, and the usage limits that keep the AI features affordable. Vercel hosts the site, deploying automatically from GitHub on every change. The Anthropic API (Claude Opus) powers the in-app assistant — which can see your plan, explain how any figure was derived, and make changes you confirm — and the spreadsheet reader that maps a file of any layout onto the planner.
One rule is deliberate: the numbers are guidance, not advice. The tool shows you the maths and the assumptions behind it and leaves every decision with you — for choices like transfers, annuities or defined-benefit options, it points you to a regulated adviser. The marginal running cost is a few pounds a month.
Built with AI: the projects
The same approach, applied to other problems. This list grows.
Personal retirement-date modelling, built in the open.
This site. A free UK retirement planner that answers the question everyone asks — when can I actually afford to stop working? It models pensions (State, workplace, final-salary), ISAs, savings, investments and property month by month with UK tax built in, finds your earliest retirement date and sustainable income, stress-tests the plan with Monte Carlo, and lets you save and compare scenarios. An AI assistant can explain any figure and edit your plan on request, and you can upload a spreadsheet for the AI to map. The projection engine runs entirely in your browser.
A daily AI news publication that largely runs itself.
Every morning, unattended, it gathers several hundred items from news feeds and podcast transcripts, drafts the day’s briefing, newsletter and podcast script, and queues them for human review. On publish it records the podcast episode with chapter marks, sends the newsletter, and serves the edition in ten languages. Its data desks refresh themselves daily: model benchmarks, an AI markets tape, and a global infrastructure tracker. Built and operated by one person directing AI tools; no development team.
True Bricks
LiveThe true total cost of owning a UK home.
A property research tool that calculates the full cost of ownership for any UK residential property: mortgage, energy, maintenance, risk factors and area data, drawn together into a single comparable view with street-level and solar-potential analysis. Built on a vibe-coding platform with an AI pair: planning and task specifications written in conversation with Claude, then executed autonomously by Claude Code against a gated test-and-verify workflow.
When Can I Retire? provides information and guidance only — it is not financial advice. See the methodology page for every assumption, source and known simplification.