Decide today. Know when to look again.

Prolepsia is a forecasting engine for teams. You describe a decision, or ask a question with a date it resolves on, and attach your own evidence. Prolepsia returns a probability or a forecast range, with the reasons and the sources it read, rather than a paragraph. A forecast can refresh on a schedule or when a condition you set is met, and it can be scored against what happened once an outcome is established.

You describe the decision, the options and what matters, including any limit an option must not cross. Prolepsia researches the uncertainties, shows how the evidence bears on each option, and points out what still needs checking. You record the choice yourself, with a condition for when to review it. The choice is yours: Prolepsia does not make it or carry it out. The app has an example decision that opens without an account.

An LLM with a search tool answers once, in prose, and is never graded. Prolepsia returns a probability or a forecast range with reasons, keeps the evidence that produced it, can refresh when a schedule or a condition you set says so, and can be scored once an outcome is established, with Brier score and log loss. A probability can be checked after the fact. A confident paragraph cannot.

Prolepsia does not require a clean warehouse. You can upload files, add notes, or connect a remote source over MCP. A database needs an MCP server or some setup before it can be used, and some direct connections are still on the way. You label what each source is and how far you trust it; answers stay traceable back to the page or row they came from. Your own sources are only half of it — Prolepsia also reads the public record around them, because the evidence that moves a forecast is usually evidence you do not own.

Yes, and that is where most of the movement comes from. Internal data records what already happened to you; it has no opinion on a rate decision, a competitor's filing, a tariff schedule or the weather. Prolepsia reads outside sources alongside your own — filings and registries, price series, official statistics, and what the market already thinks — and labels each one the same way: what it is, when it was published, and how far to trust it.

For a historical run, Prolepsia restricts its tools to sources that support a cutoff date, so the run does not read what was published after it. How well this works depends on the source: not every source keeps the figures as they stood on the day, and uploaded data can reach you later than the dates in its rows, so its publication date matters. Check the source’s limits before relying on a historical run.

On a schedule you choose, or on a condition you set, such as a filing appearing or a number passing a threshold. Both are available. Conditions are checked periodically, not the instant something happens, and checking uses credits.

A forecast can be scored once its outcome is established, using Brier score and log loss. Some questions cannot be scored, for example when no outcome is established or the question is annulled, so any record needs its coverage and exclusions stated. Prolepsia’s track record at prolepsia.com/track-record scores its resolved public yes/no and multiple choice forecasts with the Brier score, counts only forecasts made before each question resolved, and lists what it leaves out and why.

Yes. Prolepsia publishes forecasts on real questions at prolepsia.com/forecasts. Each one shows the probability Prolepsia gives today and the date it was made, the report explaining how Prolepsia reached it, the sources it read, and every earlier forecast on the same question. A forecast opens in full in the Prolepsia app, without logging in. These are Prolepsia’s own forecasts, not customer results.

Yes. Anyone can create a Prolepsia account at app.prolepsia.com/signup. A new account starts with 50 free credits, which pay for forecasts in the app. Prolepsia shows the research cost before forecasts begin. What each step costs in credits is on the app’s pricing page at app.prolepsia.com/pricing.

Yes. Prolepsia exposes its forecasts to agents over MCP (Model Context Protocol). An agent starts a forecast with one tool call and reads the result when it is ready, usually within an hour. It receives a probability or a forecast range with the reasons, instead of acting at speed on an unmarked guess.

In your own systems if you connect them. Prolepsia can send a message when a review is due, call a webhook into your own services, or answer an agent's tool call. A person records the decision. Any action that follows is built by you on the API and webhooks; Prolepsia does not carry it out.

Yes. The Prolepsia Telegram bot, @prolepsia_bot at t.me/prolepsia_bot, posts each finished forecast into a chat: the question, the probability Prolepsia gives and the report explaining it, with a link to the full forecast in the app. Start the bot in any chat, send it a Prolepsia API key, and it posts that account's forecasts there as they complete.