Only a few years ago, it would have been impossible to picture this scenario: A lead developer once gave 10 new hires at Block a codebase they’d never seen, access to AI agents, and a day and a half to solve problems that customers had brought to them in person. It was their first day, but they all delivered.
For teams wondering where junior developers fit in, this experiment raises a question. What might experienced engineers learn from them?
“It is a rollercoaster ride,” Angie Jones, the lead developer and now vice president of the Agentic AI Foundation, told The New Stack’s Frederic Lardinois at last month at AGNTCon + MCPCon North America in San Jose. Their conversation reveals a tension familiar to anyone who’s learned a new coding workflow only to find it out of date by the next model release.
Jones has led AI enablement at Block and helped maintain the open source coding agent Goose. From both experiences, she distinguishes between learning to work with an agent and learning to build one others can depend on.
A baseline developers can use to build
Jones teaches using the Model Context Protocol (MCP) to connect agents to applications and uses Agent2Agent (A2A) as an example of a standard for coordination between agents. With multiple companies agreeing on how to tackle common problems, developers have interfaces they can build around as the tools evolve.
“You can bet on it and say, ‘I can build on top of this now,’” Jones told TNS.
That still leaves plenty of engineering to do. A developer building an agent needs to understand the loop that lets it choose an action, examine the result, and decide what to do next. Someone has to decide where to constrain that loop and what access the agent should have to customer data.
The foundation’s MCP Associate certification lets developers demonstrate their understanding of MCP. At the same time, its working groups tackle issues that surface when agents move into everyday use, including identity, reliability, security, and governance.
At Block, Jones approached adoption by asking a small group of engineers to make the development environment easier for agents to use. Baking that work into the system meant developers could point an agent at a repository and benefit from what the group had learned – even if they’d missed last week’s lesson in prompting, for example.
Who gets to decide what ships?
Goose offers a maintainer’s view. Coding agents have made it easier for people to propose a fix or a feature. Each pull request still takes someone’s time to review. The project shifted to an issue-first process so contributors and maintainers could agree on the problem and the design before anyone generated a patch. “Let’s talk about the problem and the direction that we want to take with the maintainers, then collaborate on the design,” Jones said.
“Let’s talk about the problem and the direction that we want to take with the maintainers, then collaborate on the design.” – Angie Jones
Goose’s maintainers also use agents to triage issues, review incoming code, and run security scans. The tools that make it easy to open a pull request can help maintainers manage the queue, while issue discussions give contributors a place to explain why a change belongs in the project.
“A new hire may know how to get an agent moving quickly. An experienced reviewer can help them see which of those changes are worth shipping.” – Angie Jones
Jones sees a similar exchange between junior and senior engineers at work. “A new hire may know how to get an agent moving quickly. An experienced reviewer can help them see which of those changes are worth shipping,” Jones said.
Watch the full interview on The New Stack’s YouTube channel for Jones’ account of that first day at Block, what Goose learned from a flood of AI-assisted contributions, and why building agents may soon become part of the job for developers across the stack.
The post “It is a rollercoaster ride”: Angie Jones of the Agentic AI Foundation on keeping up with AI agents appeared first on The New Stack.

