AI agent time tracking
Track the work your agents do.
Your team runs Claude Code, Cursor and Copilot every day. Your tracker records the hours a person sat at the keyboard and nothing else. WhoWorked puts human hours and AI agent contributions on one timesheet, with the model, tokens and cost attached to each session.
- Agents report over MCP or REST
- Unlimited agents on every plan
- 30 days free, no card
One day, one timesheet
Human hours and AI agent contributions on the same record
The blind spot
Your timesheet has no field for most of what your team now ships.
Time tracking was designed when only people did the work. An agent that runs for forty minutes, burns 400,000 tokens and produces a working migration leaves no trace in it.
The hours keep going down and the record keeps getting thinner, at exactly the point where you need it to explain what happened.
What a conventional tracker holds
Three of the four rows are the work your team is increasingly doing. The timesheet has no field for any of them.
What goes missing
Three things your team produced last week that nothing recorded.
The session itself
An agent ran, finished a task and closed. Duration, model and outcome existed for the length of a terminal window and then did not.
What the run cost
Token spend lands on a provider invoice at the end of the month, detached from the client, the project and the deliverable it paid for.
Why the hours dropped
Delivery got faster and the timesheet recorded only the fall. The amplification that caused it is the part nobody can show.
What lands on the record
Every agent session, attached to the person who owned the delivery.
Agents report their own work while it happens. A session carries the model that answered, how long it ran, the tokens it used and what that cost, and it sits under the person who owned the delivery rather than floating on its own.
- Capture sessions from Claude Code, Cursor, Copilot, Windsurf and in-house agents.
- Keep model, tokens and cost on the session instead of on a provider invoice.
- Read the leverage ratio, which is effective output divided by human hours.
- Run projects, clients, reports and CSV export the way you do today.
Agent session
Payments API · Northstar
Model
Claude Opus
Duration
48m
Tokens
412k
Cost
$6.21
Attached to Priya K., who owned the delivery.
See a week of agent work on one timesheet.
Connect the agents your team already runs and watch the sessions arrive against real projects. 30 days, no card, and your Clockify history comes with you.
How agent time reaches the timesheet.
01
Connect over MCP or REST
Anything that speaks the protocol reports in, so a new agent your team adopts next quarter does not need a new integration built for it first.
02
Sessions arrive as they happen
The agent reports what it ran, for how long, on which model and at what cost. Nobody reconstructs the week from memory on a Friday afternoon.
03
Hours and sessions share a record
Human time is logged the usual way and lands on the same project. One timesheet holds both halves, ready for reports, invoices or a client conversation.
FAQ
Start counting all the work.
30-day free trial. Bring the time history you already have.