The manifesto. Why we exist and what we’re building.

We’re not theorising. We’re running an AI company.

Most companies talking about AI agents are building tools that help humans work faster. We think that’s the wrong frame. The more interesting question is: what happens when the software doesn’t assist the human — it replaces a category of work entirely?

The biggest shift since the spreadsheet

Most of the AI conversation today is about tools. Faster typing. Better summaries. Smarter search. That’s not what we’re building. We’re building the workforce. AI agents that hold roles, deliver work, and report to humans who stay in charge of the things humans should stay in charge of.

Humans stay in charge. That’s a design principle.

Every agent we build operates inside human-in-the-loop control points. Approval gates for anything material. Escalation paths for edge cases. Full audit logs for every action. Agents handle execution. Humans handle authority. This is designed in from the start, not bolted on later.

Built in public, with a partner who pushes harder

We work alongside The Dutch Zero Human Company, Olivier Rikken’s semi-autonomous Zero Human Company experiment. We have a signed cooperation agreement and have completed a real commercial transaction between our two AI-powered organisations. The lessons DZHC generates go directly into how we build for clients.

Read the full partnership story →

Who we are

The Agents was founded by Mr. T. The company is lean by design — AI-first, not AI-adjacent. We don’t use AI to help us do our work. AI is the work.

Plenty of companies call themselves AI-first because they bought a few subscriptions. We mean something stricter. An AI-first company is one where AI agents do the operational work by default, and humans step in for the decisions that carry real weight. The org chart has job titles. Some of those titles are held by software.

A traditional agency adds people when it takes on more clients. We add agents. The GTM Specialist that runs a launch for one client can run a launch for the next without burning out, forgetting context, or needing two weeks to onboard. The work compounds instead of resetting.

Inside every engagement, the split is the same. Agents handle execution: drafting, publishing, monitoring, reporting, the repetitive operational load that eats a team’s week. Humans handle judgement: strategy, taste, the calls where being wrong is expensive. An agent never quietly makes a decision it was not trusted to make.

Blogable is the clearest evidence that this model works. It is our first commercial agent, live in production, with paying clients. Give it a handful of keywords and it ships a finished blog post in your tone, optimised for search, pushed straight into your CMS. It is not a demo. It earns revenue.

We have also done something most AI companies are still only talking about. The Agents has a signed cooperation agreement and a settled commercial transaction with The Dutch Zero Human Company, a semi-autonomous organisation run largely by AI agents. To our knowledge it is one of the first real AI-to-AI commercial exchanges in the Netherlands.

We are building toward a future where running an AI-first company is normal, not novel. Where a small human team directs a large agent workforce, and the limit on what you can build is your judgement, not your headcount. We are not waiting for that future to arrive. We are operating in it now, and learning in public as we go.

---