My services
Hands-on Fractional Chief AI Officer
I work with your leadership team to set the AI strategy and decide where to invest. I work alongside your teams to rethink how work is shared, train them and build the tools they need. We test the changes on real cases and measure their impact on service quality and workload.
Request a meetingTell me about your project. I’ll get back to you to arrange a conversation.
01 / Compare the services
| Criteria | Decide AI audit and roadmap | Train AI training & coaching | Integrate AI agent integration |
|---|---|---|---|
| Who | Leaders and product or engineering managers. | Developers, product practitioners and small business owners. | Teams that build software with AI agents. |
| When | Several AI opportunities, unclear priorities. | Train a team on its own cases or join an existing course. | Agents are in use and checks are scattered. |
| Goal | Choose where to invest and what to test first. | Use AI on a task, check the result and repeat the process. | Verify changes before accepting them. |
| Deliverable | Audit report and prioritised roadmap. An audit Brain where included in scope. | A workshop deliverable and a method you can use on your own. | A tested setup, verification scenarios and a handover. |
| Pricing | Custom quote |
| Custom quote |
Decide
Train
Integrate
One-off engagement. We agree on the scope and deliverables for a specific need.
Ongoing support. As a Fractional Chief AI Officer, I work with your leadership and teams to follow decisions through to implementation.
02 / In practice
Adopting AI changes how teams work together. For a customer request, we rethink who does what, what an agent can take on and how the team verifies the result. People lead the thinking, design the system and make the decisions.
Before
Support forwards the message. Product asks for context. Engineering restarts the investigation.
After
Agents gather the facts and prepare the fix. The team checks the result and decides on exceptions flagged by the orange arrow.
Choose a marker on the “After” image to explore the factory.
Server room
Inference is when a trained model processes a request and produces a result. The team chooses where to run it, locally or through a provider, based on data protection, response time and cost.
Workshop
The harness gives agents context, tools and checks. For a customer ticket, they gather the facts and prepare a fix the team can verify.
Glass-fronted offices
Teams design the system, rethink their roles and practise on their own cases. They own the checks and decide how to handle exceptions.
I follow 1 task with the people who do it. We identify where work gets stuck and decide what to test first.
We equip a step with the right access and checks. We measure output quality and cost on the team’s own cases.
The team reruns the workflow and its checks on its own cases. It takes ownership of the tools and knows when to step in.
Yes. An audit helps you decide where to invest. If the need is already clear, we can start directly with training or integration.
03 / Background and evidence
I was a CTO and VP Engineering and started working with AI in 2017, in NLP and machine learning. Today I design the context and checks agents need, and help teams put them in place.
See my backgroundI joined Patrick, RTK’s creator, early on and contributed heavily to the code, including pnpm and Vitest support. RTK trims the command output that agents read.
I publish the methods and checks I use. In From Code to Merge, I show how I verify a change before merging it.
Public mentions link back to the people who cite my guides and tools. At Dev With AI, I showed how I check code written by an agent. GenAI Days invited me to walk through a real pull request review.
