With KubeCon London last week and Google Next this week, it’s been the most jam-packed 14 days of the year so far. 100s of conversations with platform engineers, leaders, vendors, students, and more.

I talked about platform engineering certifications, I talked about PlatformCon, but I also talked about something that is top of mind for everyone at Google Next this week. AI platform engineering.

AI powered platform engineering, and platform engineering empowering AI will be the defining feature of 2025. I got the chance to watch my friend Kaspar von Grünberg demo Canyon AI to dozens of people this week - the excitement for AI powered platform engineering is palpable.

Canyon is a new AI product focused on driving developer self-service through secure, context-aware automation by integrating large language models with your Internal Developer Platform to more effectively deliver on all the benefits of platform engineering.

Simply:

1. User inputs a request in natural language (create an ephemeral environment, I need permission to see ABC app etc)
2. LLM interprets the request, and can select from a set of tools provided by Canyon using the Model Context Protocol
3. Canyon offers a defined set of paths depending on their org and RBAC context
4. Canyon checks permissions to ensure the user is allowed to do what the LLM is proposing
5. If all good, action is executed

Why do I think this is so impactful? Because it takes the benefits of platform engineering (dev self service, lower ops overhead), and accelerates it with AI in a safe and secure way.

It’s the perfect example of the combination of two tech revolutions that are rocketing through the enterprise atm.

And, we are still very much at the beginning.

That means there is still a lot of learning to go.

• Answer the State of AI Platform Engineering report survey
• Test Canyon and share feedback

Let’s do this.