Meet your AI team: the 12 agents you get on day one

By Giovanni Brees, founder · Published 16 July 2026

Every page on this site says "hire your AI team." Fair question: who exactly turns up? This is the roster, agent by agent, written by the people who built them. If you want pricing, that's a separate post. This one is about what the team does.

How the team is organised

You don't manage 12 agents individually. You talk to two of them, and they run the rest.

You Atlas · CEO plans, assigns, checks, reports Gio · Specialist projects anything end to end Jack · Leo · Mia · Ruby · Max · Sofia · Clara · Penny · Victor · Ana
Brief Atlas with a goal, or hand Gio a one-off. Either way the specialists do the lane work.

Atlas is the CEO. Give your workspace a goal ("get us more plumbing jobs in Antwerp") and each morning Atlas decides what matters, briefs the right specialists, checks their output, and puts a summary on your desk at 07:00. If you'd rather steer things yourself, you can skip Atlas entirely and brief agents directly.

Gio handles specialist projects. He's the one you hand a task that doesn't fit a department: "figure out why our checkout drops on mobile", "build me a price comparison of our three suppliers", "turn this messy folder into a client proposal." He works end to end and pulls other agents in when a job is bigger than one pair of hands.

The specialists

Each of the other ten owns one lane. In the app they have their own identity, toolbelt, memory, and schedule.

If none of the twelve fits, describe the role you need in a form and the platform builds a custom agent with the same tools and the same rules.

The features that matter in week one

Verified completion. The single biggest complaint about AI agents in every review section is some version of "it said done, and it wasn't." So on KentoHQ, "done" has to be proven: each task carries checks (the file exists, the page returns 200, the draft is in the outbox) that run before you're told it's finished. Failed checks send the work back. Here's how that works in detail.

Approvals. Anything risky or outward-facing (sending an email, publishing a page, spending budget) waits for your yes. You approve from the dashboard in one click, and you can loosen the leash per agent as trust builds.

The 07:00 briefing. One page each morning: cash, competitors, customers, and the one thing worth doing today. It's the feature owners tell us they'd miss most, and it's how most people catch what the team did overnight.

Routines. Any task can repeat on a schedule you set in plain English ("every Monday at 8, send me a pipeline summary"). Agents can propose routines too, but they only start with your consent.

Meetings. A notetaker joins your calls, and the follow-ups actually get done rather than transcribed and forgotten.

Memory. Correct an agent once and it sticks. Drop in an existing CLAUDE.md, AGENTS.md, or a folder of docs from another tool and the whole team starts with your context instead of asking you to re-explain.

Where to start

Don't try to run all twelve in week one. Pick the lane where you lose the most hours, brief that one agent well, and review its first few tasks carefully. Most owners start with Jack or Mia, add Clara once they trust the tone, and only then hand Atlas the keys. The whole team is included from day one either way, free while we're in early access, so the order is yours to choose.

Hire your team - free → or see everything included.