We're not waiting for the AI roadmap. We are the roadmap.

We're not waiting for the AI roadmap. We are the roadmap.

I've spent the last few months doing something that felt equal parts exciting and uncomfortable: committing to AI across the whole company with no fixed plan, no predetermined success metrics, and no ceiling on where it might take us.

Here's what that looks like in practice.

Training the whole team

Every person in the company has had AI training and has access to AI tools. Not a lunch-and-learn. A real investment in building capability from the ground up, backed by putting the right tools in everyone's hands from day one.

My dev team didn't just attend sessions. They built agents. Each developer has taken ownership of a workflow problem and engineered a solution for it. The learning is happening in the work itself.

Agents already live

We've released two agents that are making a real difference.

The first reviews statements of work. It identifies gaps, flags ambiguity, and helps developers understand the wider impact of proposed work across our application. Before an engineer writes a line of code, they have better context. That matters.

The second is embedded in our service desk. It reviews incoming tickets, assesses the relevant application code, performs root cause analysis on faults, and guides our team on how a work order might best be fulfilled. It's not replacing expertise. It's amplifying it.

No roadmap. No ego. Just curiosity.

We haven't settled on success metrics because we're not sure yet what success looks like. And that's intentional. Value could show up in speed, in quality, in team confidence, in customer experience. We're watching for all of it.

What we do know is this: we're aiming to deliver in three areas. For our people. For our customers. For their end users.

If AI makes lives easier in any of those directions, that's progress.

What I'm seeing

The output from the team has been genuinely surprising. The quality of thinking, the pace of experimentation, the willingness to try things and iterate fast.

We're not following best practice. We're building it.

This is just the beginning.