Practical AI

Angie Jones on why agent standards need a neutral home

Angie Jones· VP of the Agentic AI Foundation at Agentic AI Foundation
·~43 min·English·Changelog
AgentsAI InfrastructureOpen SourceBusiness Strategy
TL;DR

Angie Jones, VP of the Agentic AI Foundation, explains why MCP, A2A and the other open agent standards moved into a neutral Linux Foundation home, and how rival companies now write the rules together at AI speed.

01Core Mental Model

Standards Shouldn't Live in One Company

<strong>Core agent standards like MCP moved into a neutral Linux Foundation home</strong> because companies hesitate to build on a protocol a single company could quietly change or kill.

If it lives with this one company, I don't know what they're going to do. I don't know if they'll like kill it.

— Angie Jones, Practical AI
Key Insight
The tell is adoption economics, not idealism: a standard owned by one vendor is a dependency on someone else's balance sheet, so neutrality is what makes a protocol safe enough to build a business on.
02Adoption Playbook

Move the 1%, and the 90% Follow

Instead of lifting 3,500 developers at once, Jones recruited about 50 champions from every critical repo and had them <strong>bake what worked back into the codebase</strong> so their whole team inherited it.

What I need them to do is build that knowledge back into the systems themselves so that your entire team benefits from this

— Angie Jones, Practical AI
Key Insight
The leverage isn't teaching people, it's teaching the repository. Encoding conventions in AGENTS.md and skills turns one champion's knowledge into a default every future agent picks up automatically, which scales where training sessions don't.
03The Stack

Five Projects That Wire Agents Together

<strong>MCP, agents.md, Goose, Agent Gateway and A2A</strong> each solve a different piece of the agent-interoperability puzzle, and each was donated by a different company.

right now the projects are MCP agents MD goose um agent gateway and then also just this week uh A2A which is Google's agent to agent protocol um has come into the foundation

— Angie Jones, Practical AI
Key Insight
Read the five together and a layered stack appears: agents.md tells an agent the local rules, MCP wires it to tools, Goose runs it, Agent Gateway polices its traffic, and A2A lets agents call each other. No single project is 'the standard' — interoperability is the whole row working as one.
04How Standards Get Made

Write the Rules Together, Then Compete

In the working groups, rival companies <strong>agree the standard first and compete on products second</strong>, defining the rules of the game before they play it.

they are defining the rules of the game and creating the game board together. And once they have that figured out, it's like, all right, now we can compete.

— Angie Jones, Practical AI
Key Insight
This inverts the usual worry that competitors can't cooperate. When the standard sits upstream of the product, agreeing on it isn't a concession, it's the precondition for a market big enough to be worth competing in.
05Speed

A Foundation That Moves at AI Speed

A neutral foundation is usually slow, but this one moves fast because <strong>its working-group members are paid innovators, not spare-time volunteers</strong>, racing to ship on the standard.

these are not volunteers doing this in their spare time. Like it is part of their job, you know, to innovate the future.

— Angie Jones, Practical AI
Key Insight
Speed here is a governance design choice, not luck. By staffing working groups with people whose day job depends on the outcome, the foundation aligns the incentive to ship with the mandate to standardize, the opposite of a volunteer committee.
06Global Interoperability

Global Standards Need Every Region at the Table

Because a mobile-first China, a robotics-heavy Asia and a regulated Europe all need different things, <strong>truly global standards require every region in the room</strong>.

that is why you need everyone at the table, right? And so, uh, if you only had like the your your top like fame companies in the US kind of determining all of this and there's no one from these other countries that hey, robotics is like a big deal here

— Angie Jones, Practical AI
Key Insight
Interoperability is really a proxy for representation. If only North American labs set the defaults, the standard silently encodes their assumptions; a seat for each region is how the protocol avoids being globally deployed but locally wrong.
07The Human Element

Agents Will Fade Into Everyday Life

Jones expects agents to become <strong>as ordinary as the internet and the mobile phone</strong>, while most people outside tech have barely started using them.

I do think that uh AI will become a part of you know the the everyday person's uh dayto-day just like mobile phones and internet have

— Angie Jones, Practical AI
Key Insight
The comparison to the internet and the mobile phone is doing quiet work: each went from specialist tool to invisible infrastructure only once ordinary people stopped noticing it. By that measure her bar for agent adoption isn't better demos, it's disappearance into everyday habit, which most of the world is nowhere near.