Common questions.

What is this tool?

It's an expected-value (EV) decision analysis tool. You describe a decision in plain language, and AI extracts the decision branches, suggests probabilities drawn from real-world analogues, and calculates the expected value of each path. The result is a visual decision tree you can adjust live.

What kinds of decisions work well?

Decisions with a small number of discrete outcomes and some real-world base rate — e.g. 'Should I leave my job to start a company?', 'Should I accept this offer or negotiate?', 'Should we launch in market X first?'. It works less well for decisions with continuous or deeply uncertain outcomes, or where the framing itself is unclear.

How does the AI estimate probabilities?

Claude draws on patterns from its training data — published research, historical base rates, reported outcomes — and cites the analogue it used. These are informed estimates, not actuarial calculations. You should treat them as a starting point for your own reasoning, not a ground truth.

Can I adjust the probabilities?

Yes. Every branch has sliders you can drag to override AI-suggested probabilities. Expected value recalculates live as you adjust. This is the core of the tool — the AI estimate gets you to a reasonable baseline; you calibrate it to your actual situation.

How accurate is the AI analysis?

The analysis is only as good as the framing you provide and the analogues Claude can draw on. EV calculations are mathematically correct given the inputs, but the inputs (probabilities, outcome values) are estimates. For decisions with real financial or legal consequences, use this as a structured thinking aid alongside professional advice.

Is this financial or legal advice?

No. This tool produces AI-generated structured analysis, not advice. Nothing it outputs should be treated as a substitute for professional financial, legal, or medical guidance.

What if the AI misframes my decision?

Rephrase and resubmit. The quality of the output is highly dependent on how clearly the decision is described. Specificity helps — include the relevant context, the options you're actually considering, and the timeframe.

Where is my data stored?

Saved scenarios are stored only in your browser's localStorage — nothing is saved server-side. When you run an analysis, the text of your decision is sent to the Anthropic Claude API to generate the analysis, and is subject to Anthropic's data handling policies.

Do you store or log my decisions?

The prototype does not maintain a database and does not log submitted decisions beyond what the API call requires. There is a server-side rate limiter that tracks request counts by IP address; this is not persisted.

What happens if I clear my browser storage?

All saved scenarios are deleted. There is no account system and no way to recover them. This is a known limitation of the prototype.

Why is there a rate limit on the free tier?

Each analysis makes a real API call to Anthropic. The free tier is limited to 1 analysis per day per IP address as a cost-control measure. Paid plans raise this limit significantly — see the Pricing page.

What's planned next?

Potential directions include accounts and cloud sync, shareable analyses, team collaboration, and an API. See the Pricing page for what's being considered.