FutureSearch

Description

FutureSearch is an AI forecasting platform that predicts the future and backs its work with a public track record on stocks, prediction markets, benchmarks, and forecasting tournaments. You can forecast almost any question, with costs ranging from about $0.15 to $2 per question depending on effort, and every forecast draws on a shared world model that compounds in accuracy across thousands of forecasts. Its research has been used by top AI companies including Anthropic, OpenAI, Google, and xAI.

Beyond single forecasts, FutureSearch provides a toolkit of multi-agent research teams. Its Agent Map feature runs a web-research agent for every row of a dataset, then analyzes what it finds, so you can add columns of data you do not have yet, such as enriching SaaS pricing or classifying jobs, with per-row costs and success rates shown. You can use it through the web app, a Python SDK, or by connecting it to AI tools like Claude, Claude Code, Codex, Gemini, and Cursor via MCP, skills, or plugins, typically needing only a Google sign-in.

FutureSearch offers a free start with $20 in credit and no card, plus Analyst Starter, Research, and Expert plans with better usage rates and more concurrent researchers. It is aimed at analysts, researchers, and teams who need accurate, evidence-backed forecasting and dataset research.

FutureSearch's Core Features

• Evidence-backed AI forecasting with a public track record

• Shared world model that compounds accuracy across forecasts

• Multi-agent research teams

• Agent Map for per-row web research across a dataset

• Dataset enrichment and classification as structured columns

• Python SDK and API access

• MCP, skills, and plugin integrations for Claude, Codex, Gemini, and Cursor

• Transparent per-question and per-row pricing

How to use FutureSearch?

1. 1

   Start free: sign up for $20 in credit with no credit card.

2. 2

   Ask a forecast question: pose a predictive question at low, medium, or high effort.

3. 3

   Connect your AI tool: add FutureSearch to Claude, Codex, Gemini, or Cursor via MCP.

4. 4

   Enrich a dataset: use Agent Map to research and add columns for every row.

5. 5

   Review results: check evidence, cited sources, and per-row success rates.

FutureSearch's Use Cases

• Question forecasting
• Dataset enrichment
• AI agent forecasting
• Quantitative research

FAQ from FutureSearch

• What does FutureSearch do?

  It is an AI forecaster that answers predictive questions with evidence and a public track record, and provides multi-agent research teams to enrich datasets.

• How much does a forecast cost?

  Forecasting ranges from about $0.15 to $2 per question, from low to high effort, and you are charged only for tokens actually consumed.

• Does it have a public track record?

  Yes. FutureSearch publishes results on stocks, prediction markets, public benchmarks, and forecasting tournaments, with live results updated continuously.

• How can I connect it to my AI tools?

  You can use it in Claude, Claude Code, Codex, Gemini, and Cursor via MCP, skills, or plugins, usually needing only a Google sign-in, or use the Python SDK with an API key.

• What is Agent Map?

  Agent Map runs a web-research agent for every row of a dataset and adds columns of data you don't have yet, such as pricing enrichment or classification.

• Is there a free option?

  Yes. You start free with $20 in credit and no card, with Analyst Starter, Research, and Expert plans offering better rates and more researchers.