Comparison

FlatHours vs Keito

Keito is the closest thing to us that exists. Same idea, same objection to per-seat pricing, and they got to some of it before we did. If you are comparing the two, the honest summary is that we are substantially cheaper and have a free tier, they have a longer feature list, and we take a different view of what AI is for.

Price

Read off their published pricing page on 27 July 2026. Neither of us charges per seat.

Their integrations (QuickBooks, Xero, Jira, Airtable) sit on the $199 tier. Ours are on every paid plan, because putting an export behind a $199 door is a usage fee wearing a different hat.

Where they are ahead

A comparison page that finds nothing good about the competitor is an advert. Things they have that we do not, as of 27 July 2026:

• Native Mac and iOS apps

  We have the API those would be built on, and a menu-bar app is planned, but it does not exist yet. If you want a desktop timer today, they have one and we do not.

• Idle detection that works with the app closed

  Both of us can prompt you about idle time. Theirs is backed by a desktop app, so it notices while the browser is shut. Ours only works while the FlatHours page is open, which is a real limitation.

• Planning and capacity tools

  Forecasting who is booked on what. We do budgets and burn-up, not resource planning.

• Enterprise SSO and custom roles

  SAML single sign-on against your own identity provider, on their $199 tier. We have Google, Microsoft and LinkedIn sign-in on every plan including the free one, which covers most teams, but that is social sign-in and not the same thing. If you need to enforce login through Okta or Entra, they have it and we do not.

Where we disagree about AI

Keito tracks what your AI agents cost, attributed by client. That is a good idea and they had it first.

We went a step further, because cost on its own answers the smaller question. In FlatHours an agent logs its own work through the API and the entry carries a bill rate as well as a cost. So instead of "AI cost me $340 this month", you get "AI billed $4,100 and cost $340 on this client". Whether it paid for itself is the question people actually have. There is more on how agent billing works, including what an agent actually sends and what this deliberately is not.

The other half of the disagreement is surveillance. We have an idle prompt too, added on 27 July 2026 after saying we never would, and we wrote down that we changed our mind rather than editing the old page quietly. The difference is what it can do: ours is off by default, it is a personal setting that nobody can enable on your behalf, and it can see nothing except whether a key was pressed on the page. No screenshots, no application tracking, nothing an employer can impose on staff, on any plan. Our optional AI features likewise run on your own API key, so we never charge you for tokens and nothing leaves your account unless you switch it on. Nothing here is metered either.

Where we are ahead

• A free tier that is actually usable

  One person, unlimited projects, clients and invoices, and the full Harvest import. They have no free plan and no trial.

• Agents that bill, not just spend

  Rate and cost on every agent entry, and a margin-per-client report built on both.

• Review before invoicing

  Time can be made to require approval before it can reach an invoice. They have this too; we are matching, not claiming it.

• Nothing is metered, anywhere

  No charge per invoice, project, client or task, on any plan including the free one, and no percentage of what you collect.

If you are here because Harvest repriced, the migration guide covers moving your data, and why that happened covers how to avoid picking another tool that can do it to you.