See if we fitBook a training
Harness Institute

Harness Institute is

  1. Training for engineering teams that ship with AI coding agents
  2. One shared way of working for Claude Code, Cursor, and Codex
  3. An umbrella over claudeworkshop.com, cursorworkshop.com, and codexworkshop.com

In the last two years, we have trained more than 100 engineering teams, from startups to large enterprises, Fortune 500 companies and top EU enterprises among them, to ship faster with AI while their code stays maintainable and survives an audit.


How it works

The fit check, drawn out
The discovery call, drawn out
Picking a workshop, drawn out
The training day, drawn out
What stays behind, drawn out

No slideware, no generic demos

At Harness Institute, nothing is taught from slides that could not survive contact with a real repository.

The training runs inside your own codebase, on your own tickets, with the tools your team already pays for.

The result is a way of working your engineers keep using on Monday, because it was theirs by Friday.


Who we are

Harness Institute has two cofounders. Rogier Muller in Amsterdam is CTO at a fintech and runs agents in production every day inside a highly regulated environment. Vasilis Tsolis in Athens is a partner at a legal AI company and gets regulated teams adopting coding agents without losing control.

The institute is independent and sustained entirely by the teams it trains. Our only business is training that is worth paying for. Teams report a 40% productivity increase three months after training, and rate it 5 out of 5 on average.

See more about how we decide if we fit.


Free resources

We say there is no single right way to adopt coding agents, but there is a framework. It is free, in two formats.


From the white paper

The enterprise guide to agentic development, in four lines.

  1. Review becomes the bottleneck

    When AI writes the code, the constraint moves. Teams that only speed up writing queue the same work in front of a human who now reads more than they type.

  2. Quality drifts without anyone noticing

    Nothing fails loudly. Conventions erode one accepted suggestion at a time, and the cost shows up months later in code nobody wants to touch.

  3. Adoption sticks with a few champions

    Two people get fast, everyone else watches. The licence spend is the same either way, which is why usage numbers flatter and delivery does not move.

  4. Delegate, Review, Own

    Our framework, phase by phase, plus the six practices high-performing teams use to fix all three problems above.