The Goose agent originated at Block, and the Model Context Protocol (MCP) originated at Anthropic.
“Goose came from Block and MCP came from Anthropic.”
Person
Angie Jones
1 episodes · 21 quoted claims
Angie Jones describes the Model Context Protocol (MCP) as what agents use to connect to applications and tools.
“MCP is the model context protocol. This is what agents use to connect to applications and tools”
OpenAI, Anthropic and Block wanted to form a foundation for agentic AI and recognized they should not be sole owners of the standards.
“As Openai, um, Anthropic and Block wanted to form a a foundation for agentic AI, they understood that.”
OpenAI, Anthropic and Block formed the Agentic AI Foundation under the Linux Foundation.
“And so those three companies came together to form the agentic foundation under the Linux foundation, and so Linux Foundation has been around for decades”
According to Angie Jones, the Agentic AI Foundation was formed at the end of 2025.
“this agentic AI foundation was formed, um, end of 2025, so it's been what about eight or nine months at this point.”
The foundation's projects include MCP, AGENTS.md, Goose and Agent Gateway, and Google's A2A protocol joined in the week of recording.
“right now the projects are MCP agent, SMD, Goose, um, agent, Gateway, and then also just this week, uh, A2A, which is Google's agent to agent protocol”
AGENTS.md is a standard created by OpenAI for how codebases communicate operating instructions to agents.
“agent MD. So this was a standard created by Open AI, and this one standardizes, like how projects, uh, mainly like codebases, communicate their their operating instructions”
Agent Gateway came from a company called Solo, which created it to mediate agent traffic such as MCP traffic.
“agent gateway came from a company called Solo, uh, and uh, Solo, uh, created agent gateway to mediate traffic.”
A2A is Google's agent-to-agent protocol, which lets agents coordinate with and delegate work to other agents.
“A2A is the agent to agent protocol. And this one is really cool in that this is how agents can coordinate with other agents.”
Angie Jones says that at Block she was tasked with teaching the entire company how to use AI agents.
“One of my tasks was to, uh, basically teach the entire company, twelve zero people, how to use AI agents.”
Block created an internal AI agent called Goose, which developers used to automate engineering tasks and help with coding.
“our company block created like this internal AI agent, uh, named Goose”
Block is the financial technology company that houses Square and Cash App.
“the finance tech company that, uh, houses square and cash app, right?”
Angie Jones worked with Jack Dorsey at Twitter, and he then brought her to Block.
“So I worked with Jack at at Twitter, and then he brought me over to Block when they when they started”
Jones used the 1-9-90 rule, under which about 1% of a community are creators, 9% tinkerers and 90% consumers, to organize AI adoption.
“it's called the the one, nine, ninety rule, and this rule essentially says that in any community, there's gonna be one of that community that are creators.”
Jones assembled a group of about 50 engineers from across Block to test AI techniques and embed them in their repositories.
“And so they were there with me, we drank it, it was 50 of them, um, drank it from the fire hose.”
Jones says developer attitudes toward AI coding tools shifted around November 2025, when the models improved markedly.
“But it was like this point in about November of 2025.”
Jones cites agentic commerce as an area where PayPal and Stripe need to work together on standards rather than one setting them alone.
“if we're talking about agentic commerce, for example, it PayPal doesn't want Stripe figuring that out by themselves, right?”
Jones says mobile-first markets like China need agent-to-agent protocols, and gives WeChat as an example of an app using them with various agents.
“they could, like, use that, like, for example, like WeChat uses that, uh, you know, with various agents and stuff like that.”
Jones says the EU AI Act changes the transparency required of AI systems offered to people in Europe.
“Europe is another good example, so they just rolled out the EU AI Act, right?”
The foundation's working groups hold public meetings that anyone can join.
“all of these UM working groups are open, so anyone can join them.”
Jones points to the MCP Discord server as an example of competing companies collaborating on open-source work.
“all you gotta do is go in the MCP Discord server. It's a thing of beauty.”