24hrs Inside the $300M Startup Building the Infrastructure Layer for AI Agents

In this episode, I go behind the scenes with James Everingham, Chris Waterson and the team at Guild.ai. Guild.ai is building the control layer that lets organisations deploy, govern and monitor every AI agent running inside their systems. As businesses roll out more of these agents, most lose track of how many are active, what they can access, who owns them, and what they're costing. Guild gives every agent scoped permissions, approval gates and an audit trail, so nothing runs unchecked, no matter which AI model is powering it underneath. James Everingham and Chris Waterson both spent time at Meta before joining the founding team at Guild in 2025. Between them, the founding team had already built and scaled multiple ventures, including companies that went on to be acquired by major technology companies such as Yahoo. They were joined by a team of former colleagues as they set out to build Guild’s control plane for AI agents. Guild has raised $44 million combined across a seed round and a Series A, with the Series A led by Google Ventures, Khosla Ventures and backing from NFX. The two rounds closed within four months of each other. James's framing for the problem is simple. An AI agent is like a new hire, and no company hands a new hire the keys to the business without some rules of the road. As agents move out of experiments and into production, this episode explores whether the layer that controls them ends up mattering as much as the models running underneath. Inside this episode: - James on "an AI agent like a new hire" thesis and why agents need the same accountability as employees - Howard on what an engineer's job looks like once agents are doing the work - Inside Guild's weekly all-hands, The Great Hall - The team's literature club breaking down a Microsoft case study on cutting agent costs by over 70% - A demo of Guild Insights, showing real-time spend and…