Hi, I’m Tommy. I’ve spent most of my career on platforms: building them, running them and deciding where they should go next. These days AI agents write a good part of my infrastructure code, and I spend my time on the guardrails.

Most of what I write here comes back to two questions.

How do you let AI agents do real work without handing over the responsibility?

• Making myself redundant?
• What my agents aren’t allowed to do
• Brilliant at filtering, rubbish at being you

How do you set a platform direction that actually lands in production?

• Why I’m suspicious of an empty queue

And a bit of the person behind it

• The day we got phished

If you’d rather follow along, I’m on LinkedIn and Bluesky.