Engineering Teams Ditch Traditional Structures as AI Agents Take Over Code Creation

Breaking: AI Agents Trigger Rapid Engineering Restructuring

San Francisco—Last night, at least 10 events aimed at connecting AI startups with VCs crowded the city's calendar. But one stood out: Camp AI’s “Agents at Work” event, hosted by Auth0, showcased companies already reorganizing their engineering processes around AI agents. The vendor ecosystem—including Browserbase, Mastra, Fireworks AI, Drata, Mya, MindFort, and Corridor—is racing to enable secure, performant agentic AI. Their own reorganization stories proved most revealing.

Agentic AI Reshapes Team Structures

Paul Klein IV, founder and CEO of Browserbase, captured the night's most memorable line: “If AI is not doing your whole job it’s a skill issue at this point.”

Abhi Aiyer, founder and CTO of Mastra, described dramatically smaller teams now capable of executing far larger scopes. “You can have one person run a whole feature project because they have an army of one to infinity AI agents behind them,” Aiyer said.

The New Bottleneck: Code Review

Several panelists argued that AI generates code faster than organizations can safely review and operationalize. Aiyer noted that engineering teams are opening significantly more pull requests while review throughput becomes the bottleneck.

Klein stressed throttling experimental AI output to lower risk. “If you are in the critical path and customer facing, no slop,” he said. “If you are not critical path, not customer facing, slop away.”

Trust and Ownership Challenges

Observability and accountability emerged as recurring themes. Rob Ferguson, VP of technology and strategy at Fireworks AI, argued ownership cannot disappear simply because AI generated the output. “It doesn’t matter if you typed it or prompted it, you own it,” Ferguson said.

Bhavin Shah, VP of AI product at Drata, emphasized enterprise AI's need for detailed auditability. “The agent is constantly telling the user, here is the action I’m taking, here is what I’ve done,” he said.

Securing Agentic Workflows

Auth0’s demos focused on authentication, authorization, and runtime controls for AI agents interacting with APIs and Model Context Protocol (MCP) servers. The company’s new MCP authentication product, reaching general availability this week, secures how agents interact with MCP servers and APIs.

Monica Bajaj, SVP of engineering at Okta, highlighted minimizing risk as agents operate autonomously. “How do we make sure that those tokens are not long-lived tokens,” she asked, stressing the need for short-lived, scoped access.

Background

Agentic AI—software that can autonomously perform multi-step tasks—has rapidly moved from research labs to production. Engineering teams are finding that these agents can write, test, and deploy code independently, forcing a rethink of traditional roles and review processes. The shift mirrors earlier DevOps transformations but with even greater speed.

Experts warn that organizations must adapt their hiring, training, and security practices to keep pace. The companies showcased last night represent early movers, but industry-wide adoption is expected to accelerate.

What This Means

For engineering leaders, the message is clear: reorganizing around AI agents is no longer optional. Teams that resist risk being outpaced by smaller, agent-enabled teams that deliver features faster. However, new bottlenecks in code review and security demand urgent attention.

Investors and VCs are taking notice. The proliferation of AI-agent-focused events like “Agents at Work” signals a funding wave for startups building the infrastructure to manage, secure, and audit autonomous code generators. The next year will likely see a massive restructuring of how software is built.

— Reporting from San Francisco