Keito is time tracking software built for how professional-services work is delivered now — by people directing AI agents inside harnesses like Claude Code, Codex and DeepSeek’s new one — and for how it still has to be billed: honestly. It records human hours and agent sessions (with their cost) on the same project, puts both through a review step, and invoices both as themselves, at a flat price that doesn’t change when you hire.
I co-founded the consultancy that built it, and we run the company on it. It is for professional-services firms — consultancies, agencies, engineering and data teams — with a team and clients to invoice, including the many that haven’t put a harness into delivery yet and want the billing side ready before they do. This is why it exists, what it does, what it costs, and who shouldn’t buy it.
Why we built time tracking software instead of buying it
For eight years our consultancy worked the way every services firm works: engineers billed by the day against statements of work, hours in, invoice out. Then a project we didn’t want — a fixed-fee integration in a language I don’t write — pushed us into agents out of necessity, and the unlock turned out to be context rather than autocomplete. So we gave the AI everything: every call transcribed, a repo per client holding the whole history, an hourly job pulling fresh context in.
Today requirements arrive from calls and tickets, an agentic engineering loop runs for an hour or two inside a harness — Claude Code and Codex for us, with DeepSeek’s new harness arriving — a human reviews every output before it ships, and out comes a pull request — delivered, working code, not a timesheet. Each engineer runs two or three engagements at once where it used to be one; my org chart is a couple of dozen scheduled agents, most of which run while I sleep. Humans at the edge, AI in the middle.
That model broke every time tracker we’d used, because they could only bill what a human logged on a timer. Agent work either vanished into a “misc” line or got padded into human hours — neither is honest, and clients can tell. We’d also spent years resenting per-seat pricing in a business where the number of humans was becoming the wrong thing to meter. So we stopped shopping for a replacement and built the tool we’d been describing to each other in Slack. Then we used it on ourselves before we sold it to anyone. I wrote up the operating model and the three billing decisions behind it in Time tracking software for the AI-agent era: humans at the edge, AI in the middle, both on the invoice.
What Keito does
Keito keeps the whole loop that professional-services teams actually run, not just the timer at the front of it.
Track. Timers, manual entry and a weekly grid on the web; native Mac and iOS apps; a CLI for engineers; and a GitHub Action that turns issues, pull requests and reviews into billable entries. Time is always attached to a client, a project and a task, so it’s billing evidence from the moment it’s logged.
Review. Timesheet submissions and manager sign-off, with configurable email nudges — a review step before anything reaches an invoice, which matters more, not less, when some of the work was autonomous.
Bill. Expenses with approvals, project budgets with alerts, and invoices generated directly from approved time — recurring invoices, attachments, online payment through Stripe with no platform fee, and invoice sync to Xero and QuickBooks on the Business plan.
Report. Utilisation, budget burn and profitability by client and project; team reports for managers; nine starter report templates on day one.
Connect. A REST API on every plan, Zapier through OAuth2, and an agent integration layer — MCP, Node and Python SDKs — so tools and agents can log time and expenses without a human clicking a timer.
The part nobody else does: AI agent work on the invoice
Every Keito plan includes unlimited AI agents. An Agent Skill logs Claude Code and Codex sessions as source=agent time entries; the API and SDKs do the same for anything else you run, DeepSeek’s harness included. Agent time and LLM costs sit alongside human hours on the same project, go through the same review, and appear as their own line items on the invoice — so AI-assisted delivery can be billed honestly instead of hidden in a “misc” line, and clients can see exactly what people did and what agents did. See invoicing agent work for how the lines are built.
The finding that surprised us most: clients don’t blink. Nobody has pushed back on being invoiced for agent work — they push back on opacity, on not knowing what they’re paying for. Showing the work as itself is the whole trick, and it is the axis most buyers still forget to ask about when they choose a tool.
If your firm hasn’t adopted a harness yet
Most professional-services firms are still at the “a few people use Copilot” stage. The harness changes the unit of work; the invoice has to follow, and retrofitting billing onto delivery you’ve already shipped is where the “misc” lines come from. Set up three things before the harness, not after: a way to record agent sessions as work, a review gate that treats agent output like a junior engineer’s, and an invoice line that shows the client what agents did. Keito is those three things, in that order.
What it costs — and why the model matters more than the price
Flat tiers. You pick a plan for the features you need, not the headcount you have.
| Plan | Price | Who it’s for |
|---|---|---|
| Solo | $19 USD/month | 1 licensed user, unlimited AI agents; Mac and iOS apps, invoicing, Stripe payments, API, exports |
| Pro | $49 USD/month | Unlimited licensed users; adds team management, approvals, team reports |
| Business | $199 USD/month | Unlimited users; adds Xero and QuickBooks sync, planning, SSO, priority support |
Fifty people pay the same as five. In an operating model where agents do a growing share of the work, headcount is the wrong thing to meter — and whatever tool you end up on, price the model at twice your current team, not today’s. More on why we chose this in why flat pricing matters.
Bringing years of data with you
The thing that keeps teams on software they’ve outgrown is the sunk cost of their history, so import is part of signup: connect your existing tool and Keito moves clients, projects, tasks, team members, time entries, expenses, invoices and estimates in one pass, in dependency order, while the old system stays untouched as your read-only fallback. The migration docs include a verification checklist and a cutover plan, and there is a step-by-step guide for the most common starting point.
Who should not buy Keito
Keito is built for firms with a team and clients to invoice. If you’re solo, have a handful of clients, don’t invoice by the hour and have no AI in your delivery, you don’t need to pay anyone — a free timer or a spreadsheet is fine, and I’d rather you read that here than discover it after paying us. Keito also deliberately has no employee monitoring (no screenshots, no activity levels); if that’s what you’re after, we’re the wrong tool. And it’s a younger product with a smaller integration catalogue than the incumbents.
Keito earns its keep when you have a team, invoices that need review before they go out, a bill that shouldn’t grow every time you hire, and work that AI agents now help deliver.
Time tracking software by role
The loop is the same everywhere; the details differ. See how it fits for accountants, architects, consultants, developers and agencies billing AI work, and the Harvest migration guide if that’s where you’re coming from.
Frequently asked questions
What is time tracking software?
Software that records how working hours are spent — logged against clients, projects and tasks — so a business can see where time goes, what it costs, and what to bill. Keito’s version keeps the whole loop: track, review, expense, invoice, report.
Is Keito good time tracking software for a small business?
If the business bills clients for its time, yes — that’s what it’s built for: client and project context on every entry, a review step, invoicing with online payment, and a flat price so growth doesn’t raise the bill. If you only need to know where your own hours went, a free timer is enough.
Can time tracking software track AI agent work?
Most can’t — nothing fills in a timesheet for a coding agent. Keito records agent sessions and LLM costs as their own time entries alongside human hours, reviews them, and puts them on invoices as separate lines. Unlimited agents are included on every plan.
Can I bring data from my current tool?
Yes — import runs during signup and moves clients, projects, tasks, time, expenses, invoices and estimates in one pass, with the old system left untouched until you have reconciled totals. There is a detailed guide for the most common starting point.
Is there a free plan?
No — a 14-day trial. Solo is $19 USD/month; teams are $49 USD/month with unlimited licensed users. Every plan includes unlimited AI agents.
How long does migration take?
The importer runs during signup; budget an afternoon for reconciling counts and totals against your old tool before you cancel it.
Track time for humans and AI agents in one place
Keito is the time tracking and invoicing tool we built for the way our own consultancy now works. If your team bills for work that people and AI agents deliver together, start a free trial — the importer runs during signup.