Units: People, Token & Compute

A Unit is a known quantity of work, about one person-day. You buy the number you need each month and mix three kinds to fit the job. It replaces the usual tangle of separate contracts and hourly rates with one predictable cost you control.

The three kinds of Unit

People Units are real people. Business analysts, solution architects, and data specialists for discovery, meetings, and cleaning up your data. This is the human work behind data insights and analysis and custom software.

Token Units are AI agents at work. They’re the tokens that feed our coding agents while we build, and that keep your live AI agents running each day. Under the hood that’s machine learning and generative AI doing the heavy lifting.

Compute Units are the infrastructure your systems run on: the database and platform behind your agents and your cloud setup.

How Units flow over a project

The mix shifts as the work matures. A big discovery runs heavy on People Units. Then we switch to Token Units to feed the coding agents that build it. Once it’s live, your agents run each day on Token and Compute Units, and People Units drop right back. You can watch this play out on the allocator on the Technology Partner page.

About the cost of one employee

We don’t publish a per-Unit price, because the right number depends on how much you want to get done. For most small to medium businesses, a working month of Units costs about the same as one employee. Want more done faster, add Units for that month. Want to hold steady, dial it back. There are no surprise invoices, and there’s a per-customer discount on the Unit price.

Scaling up and down

Units are made to flex. You scale the total to your goal each month and re-mix as you go. See our delivery model for how that runs in practice, and how we engage for how a plan becomes a month of Units.

Ready to model your month? Try the allocator on the Technology Partner page, or take the AI Roadmap Interview.