Works with Claude Code, Codex, Gemini CLI, and any MCP client

Data collaboration for humans, agents, and AI coding tools

Hutch is the missing glue between your agents, your people, and your processes.

Or create an account with email

What's a data collection?

A collection is a flexible container for structured records. No schema to define upfront — just start adding data.

AI AgentsWeb Forms>_AI CodingA HutchData Collection ViewsReportsAgent Queries

Example: content improvement pipeline

A content-improvements collection that connects two AI agents and your team.

OpenClaw

Step one

An OpenClaw agent crawls your website and finds errors. For each error, it adds a record to the collection.

In the OpenClaw prompt:

For each page with an error found, add the URL and a short description to the issues collection in Hutch.

Step two

The collection is published as a live web page. You review the suggestions, edit if needed, and mark the ones you like as approved.

OpenClaw

Step three

A second OpenClaw agent queries for approved records, picks them up, and applies the changes — updating page content, rewriting meta descriptions, and marking each one done.

Step four

Anyone on the team can submit pages they've noticed need work via a web form on the same collection. Same queue, same review process — agents and humans side by side.

Track agent actions

Every action your agent takes becomes a queryable record. See exactly what happened and when.

Add human checkpoints

Agents propose, humans approve. Build approval queues without writing a single line of UI code.

Combine human + agent data

Web forms and API calls feed the same collection. One dataset, every source.

Get reports without building dashboards

Published pages, views, and aggregation queries. Share a URL with your team.

Built for real workflows

Lead pipeline

Agents find prospects online. Your team adds leads manually. Everything in one place, ready to act on.

Agent audit trail

See every action your agents took, filterable by date, type, and status. Share with your team via a link.

Approval queue

Agent drafts work, you review it on a web page, agent picks up approved items. Human-in-the-loop without the plumbing.

Multi-source reports

Data from forms, APIs, imports, and agents. Query and aggregate without building a dashboard.

What powers it

Schema-optional JSONB storage·Auto-creating collections·Upsert via unique_key·Full-text search & aggregation·Free to use