.png)
One system for humans and agents.
Hatch reads the work already happening across the tools your company runs on, and shows who is performing, where capacity sits, and what the agent layer is actually returning. No forms, no surveys, no migration.
Four things you can finally see.
Read the work, not the reporting on it.
Hatch connects to Slack, GitHub, Linear, Asana, Close CRM, Rippling, Greenhouse, HiBob and Google Workspace, plus agent activity over MCP. It reads the work itself, so nobody has to fill in a form describing what they did.
- Read-only access to the tools your team already works in
- Human and agent activity land in the same record, not two systems
- Personal extensions people opt into: Fathom, Granola, Fireflies, Google Calendar, Whoop, Oura
Measure what agents cannot carry.
Judgment, context, point of view and taste do not show up in a ticket count, and they are the part of the work an agent cannot take over. Hatch reads them from the work itself across eight behavioral dimensions.
- Eight behavioral dimensions, applied to the work people actually did
- Grounded in nine peer-reviewed meta-analyses and 2,200 behavioral profiles
- No questionnaires and nothing for anyone to fill in
-- read-only view of work already happening CREATE VIEW work_by_actor AS SELECT bucket(ts, '1 week') AS week, p.team, sum(p.minutes) / 60 AS hours, count(DISTINCT s.person_id) AS people FROM work.actions p JOIN work.people s USING (person_id) WHERE p.actor_type = 'agent' GROUP BY 1, 2;
Every agent action, attributed.
The agents came with a bill. Hatch reads agent activity over MCP, including Anthropic and OpenAI, and attributes every action back to the person, the team and the piece of work it was running for. You see what the agent layer returned, not how many seats were switched on.
- Every agent action tied to a person, a team and a piece of work
- Agent activity read over MCP, including Anthropic and OpenAI
- Adoption is not the measure. Returned work is.
Everything else you would be guessing at.
Capacity view
Where output is constrained and where it is idle, by team and by person.
Work provenance
Every number traces back to the work that produced it, in the tool it happened in.
Historical baseline
Compare this quarter against the record of work already sitting in your tools.
Role-based visibility
Founders, finance and managers each see the slice they are accountable for.
Manager alerts
A note when a team's output or capacity moves enough to be worth a conversation.
Weekly digest
One email a week: what moved, where capacity sits, what the agent layer returned.
Built on published research.
Published work, cited rather than claimed.
The base the dimensions are read against.
Applied to work people already did, not to a questionnaire.
Find out what you actually have.
Thirty days, one named engineer, and a readout of what your team and your agents actually produced. $2,500, credited in full against your first year.