futuresearch

FutureSearch Python SDK

Deploy a team of researchers to forecast, score, classify, or gather data. Use yourself in the app, or give your team of researchers to your AI wherever you use it (Claude.ai, Claude Cowork, Claude Code, or Gemini/Codex/other AI surfaces), or point them to this Python SDK.

Requires Google sign in, no credit card required.

Quick installation steps:

Claude.ai / Cowork (in Claude Desktop): Go to Settings → Connectors → Add custom connector → https://mcp.futuresearch.ai/mcp

Claude Code:

claude mcp add futuresearch --scope project --transport http https://mcp.futuresearch.ai/mcp

Then sign in with Google.

Operations

Spin up a team of:

Role	What it does	Cost	Scales To
Agents	Research, then analyze	1–3¢/researcher	10k rows
Forecasters	Predict outcomes	20-50¢/researcher	10k rows
Scorers	Research, then score	1-5¢/researcher	10k rows
Classifiers	Research, then categorize	0.1-0.7¢/researcher	10k rows
Matchers	Find matching rows	0.2-0.5¢/researcher	20k rows

See the full API reference, guides, and case studies, (for example, see our case study running a Research task on 10k rows, running agents that used 120k LLM calls.)

Or just ask Claude in your interface of choice:

Label this 5,000 row CSV with the right categories.

Find the rows in this 10,000 row pandas dataframe that represent good opportunities.

Rank these 2,000 people from Wikipedia on who is the most bullish on AI.

---

Web Agents

The base operation is agent_map: one web research agent per row. The other operations (rank, classify, forecast, merge, dedupe) use the agents under the hood as necessary. Agents are tuned on Deep Research Bench, our benchmark for questions that need extensive searching and cross-referencing, and tuned to get correct answers at minimal cost.

Under the hood, Claude will:

from futuresearch.ops import single_agent, agent_map
from pandas import DataFrame
from pydantic import BaseModel

class CompanyInput(BaseModel):
    company: str

# Single input, run one web research agent
result = await single_agent(
    task="Find this company's latest funding round and lead investors",
    input=CompanyInput(company="Anthropic"),
)
print(result.data.head())

# Map input, run a set of web research agents in parallel
result = await agent_map(
    task="Find this company's latest funding round and lead investors",
    input=DataFrame([
        {"company": "Anthropic"},
        {"company": "OpenAI"},
        {"company": "Mistral"},
    ]),
)
print(result.data.head())

See the API docs, a case study of labeling data or a case study for researching government data at scale.

Sessions

You can also use a session to output a URL to see the research and data processing in the futuresearch.ai/app application, which streams the research and makes charts. Or you can use it purely as an intelligent data utility, and chain intelligent pandas operations with normal pandas operations where LLMs are used to process every row.

from futuresearch import create_session

async with create_session(name="My Session") as session:
    print(f"View session at: {session.get_url()}")

Async operations

All ops have async variants for background processing: