Building the Foundation for Agentic AI

I’ve spent my career helping engineering teams build and ship systems that scale. That perspective shapes how I look at every new wave of technology.

Agentic AI is moving fast, and we’re at a point where demos won’t cut it anymore. Developers are already building with agents in ways that would have felt out of reach even a year ago. They’re creating agentic workflows across different tools in production environments, trying to stitch together something that actually works.

The problem is that every agent behaves a little differently, so developers are feeling the pain of reconfiguring their MCP servers, skills, memory, and context over and over again. It’s clear that this needs to be more consistent and reliable.

That’s what led me to join the Agentic AI Foundation (AAIF).

AAIF brings together the teams building today’s AI systems to work on the shared infrastructure behind them. The goal is to make it possible for agents to work across tools and environments without constant rewiring. Under neutral governance, that work centers on open standards and protocols that enable interoperability.

There’s already meaningful work happening across the ecosystem. Projects like MCP, goose, and AGENTS.md are shaping how developers operate agentic systems today.

From my experience leading AI adoption in complex production environments, I know how critical it is for developers to have systems they can trust. As agents become part of everyday workflows, understanding how they behave when integrated into existing systems becomes essential.

In the near term, my focus is on making this practical. That means publishing working implementations that show how to build reliable agentic workflows using the same standards we’re asking others to adopt. It also means working closely with developers to understand where friction still exists and translating that into improvements that make them more effective. And it means ensuring that the foundation demonstrates how these standards enable real progress.

Success will show up in how the ecosystem moves. More tools supporting these standards out of the box, more integrations across vendors, and interoperability becoming a default expectation.

Most importantly, it will show up in the strength of the ecosystem that forms around this work. Open source only works if people actually use it and build on top of it. That’s the goal here. Not just defining standards, but seeing them adopted and improved in practice.

I’m excited to build alongside this community and help shape what comes next.