ClueCon Weekly with Adam Kalsey: Running Autonomous AI Agents in Production

This week on ClueCon Weekly, Luca sits down with Adam Kalsey, VP of Product at SignalWire and longtime AI learner, to talk about what it actually looks like to run an autonomous AI agent in production. Not as a demo. Not as a toy. As a real, working team member.

Adam shares the story of building Poppy, SignalWire's AI-powered product operations agent — including the brutal early lesson of running Opus with extended thinking until it burned through $40/hour in tokens, the model tiering strategy that brought costs down to under $20/day, and how he's architecting for real-world constraints like credential management, blast radius, and access control.

Also covered:

• Git-based CI/CD for markdown "soul" files

• Docker isolation

• Self-reinforcing skill loops

• The surprisingly thorny problem of keeping an agent from accidentally mass-emailing your customers

This is the episode for anyone who wants to move beyond "I ran Claude on my laptop" and into something that actually runs overnight.