Enterprises are being sold AI agents faster than anyone has agreed what a good one looks like. The MACH Alliance workgroup I co-chair exists to slow that conversation down to the speed of getting it right.
We work on the questions that sit underneath the demos. How do agents operate across a composable architecture instead of a single monolith? Where are the trust boundaries? What does adoption look like when the buyer is an enterprise with procurement, security review, and a platform team that will still be there in three years?
Guidance, not another framework
The output is shared vocabulary and guidance that vendors and enterprises can actually build against. Not a new protocol. Not a vendor bake-off dressed up as a standard. A way to talk about agents in a MACH stack without pretending the CMS, the commerce engine, and the search index are one brain.
I took the co-chair role because I would rather help shape the standards than complain about them later. It is ongoing work under the MACH Tech Council, and it pairs directly with the agent and MCP projects I build on my own time. The workgroup tells me what the enterprise needs. The code is where I go to find out whether it is possible.