Skip to content

field notes · ai and agents

AI training

AI trainingAn isometric drawing of a six-week AI training course as three bays. On the left an audit board with four lines to tick off in week one. In the middle a bench marked your code, where piles of exercise cards build up in weeks two and three. On the right a board of twelve tests and a marked footprint where the team builds one feature, a block a week, from week four to week six. Three people stand at the front and walk to the bay each week is in, and a binder for the handover waits at the front right.week 1weeks 2 to 3weeks 4 to 6your code123456

Six weeks, left to right: an audit, exercises from your own code, then one feature your team builds. Run a week.

Six weeks, and your team ships an AI feature it can test.

Book a free first callEstimate the cost

In short

AI training is a six-week programme that teaches a product team to build with language models: writing prompts, testing output, choosing models and shipping one real feature. Zenoware, a small studio, runs it for developers, product managers and testers, in person or remotely. Week one is a readiness audit of your systems and rules. The exercises come from your own code, and the final three weeks build a working assistant or workflow. The first call is free.

What you get

  • A readiness audit you can act on in week one

  • Prompts and tests your team wrote, in your repository

  • One assistant or workflow, built by your own people

How it runs

  1. 01

    Audit

    Week one: we check your systems, rules and habits, and agree a starting point.

  2. 02

    Write the exercises

    Weeks two and three: exercises cut from your own repositories, one set per role.

  3. 03

    Build one thing

    Weeks four to six: small groups build an assistant or workflow, with its tests.

  4. 04

    Hand over

    Your leads keep the guardrails, the test suite and a plan for what's next.

Seen in

Who you'll work with

Brush sketch of JP Caparas

JP Caparas

Founder and lead engineer

Thirteen years of principal engineering. OpenAI Codex and Devin ambassador in New Zealand. You talk to the person who writes the code.

About the team

Questions

What is AI training for a software team?

A hands-on course where the people who build your product learn to work with language models. That means writing prompts that hold up, testing output the way you'd test code, picking a model for the job and knowing what it costs. Ours runs for six weeks and ends with something your team built and can keep running.

Who is AI training for?

Developers, product managers and testers. A group is usually up to eight people, and larger groups are possible. Each role gets its own exercises. A product manager doesn't need to write a test harness, but should be able to read one and say what good output looks like.

What is an AI readiness audit?

A short review of whether your systems, data and rules are ready for AI work. We look at where the useful data lives, who may see it, how releases happen and what your people already do with AI tools. It's week one of the programme, and you get a written list of what to fix first.

Do you use our own code and data in the training?

Yes. Generic exercises teach generic lessons, so we cut ours from your repositories and from the questions your customers really ask, inside whatever privacy rules you set. If some data can't be used, we work from samples you approve.

What do we have at the end of the six weeks?

A working assistant or workflow your team built, a set of prompts and tests in your repository, and written guardrails for the rest of the company. Your leads also get a plan for the next thing to build, so the habit carries on after we've gone.

You might also need

Tell us what isn't working.

The first call is free, and we'll tell you straight whether we're the right people for it.

Start a project