AI Deployment Diaries brings you stories from the front lines of enterprise AI deployment. Host Parth Mehta sits down with the operators and practitioners deploying AI inside real companies on what works and what doesn't.

Travis Frisinger built a software factory that builds itself. In this episode he takes it apart: what a software factory actually is (intent, an operator, a spec, standards, a harness), how he got it to bootstrap and woke up to 30 merged pull requests, and what changes when the agent starts reading its own transcripts to learn. We get into why engineers have to stop handing their minds fully to the agent, the ego hit of watching a machine decide your technical debt was never worth cleaning up, and what custom software looks like in 2027 when a client can get an app 3D printed instead of paying for a six-figure build.

Travis Frisinger is Head of Agentic AI at 8th Light, where he leads the Agentic AI Studio. He is the creator of HydraFlow, an open-source software factory, and writes about software manufacturing at aibuddy.software.

Read Travis's essay on the ideas in this episode, "Software Factories Need Building Codes":
https://aibuddy.software/software-factories-need-building-codes/

HydraFlow on GitHub: https://github.com/T-rav/hydraflow

How to get in touch with Travis:

LinkedIn: https://www.linkedin.com/in/travis-frisinger/

How to get in touch with Parth:

LinkedIn: https://www.linkedin.com/in/parthcmehta/

X: https://x.com/mr_parthmehta

#ai #aiagents #softwarefactory #enterpriseai