Software engineer building and experimenting with AI agents.
AI has changed how software gets written for good, and it demands a new skill set rather than replacing the old one. You still need to know what quality code looks like — and half the job is now understanding how agents work, how to manage them, and how to evaluate them. The other half is improving the system that does the writing.
Working on
Both are experiments toward a software factory — agents that write production-quality software while the engineer's job becomes making that output good.
- acai — where I started: coding agents you work with, in the loop.
- cake — agents that run autonomously and safely with nobody watching. It takes the bitter lesson seriously: models will keep absorbing more of the harness's job, so the harness stays as thin as it can.
- cake-repl — makes cake behave like acai; it's also the interface I built this site through.
- cake-telegram — runs cake as a bot, private for now.
Background
- Since 2007 — software engineer: most of it in education software, and the last three years in enterprise and cloud.
- Now — I lead AI-for-software-development for my company's Cloud and Software Group: training engineers to use AI in their work, helping define our AI development lifecycle approach, administering our AI coding tools, and writing and maintaining internal resources — MCP servers, CLIs, and prompt and agent skill libraries. I advise teams across the org on adopting AI.
- Before — AI work across education software, plus shipped SaaS products for schools and organizations.
- Education — dual master's in information science and library science.
Full history on LinkedIn.
About
I started in library science, added information science, and graduated with a dual master's. Grad school was my first brush with AI — Prolog, ontologies, RDF triples — and then I barely touched it for years, until generative AI changed what was actually possible.
Outside work I read and draw. I'm still interested in photography, but I don't get to it much anymore.
Getting in touch
Email is the best way to reach me: travis@travisennis.com. You can also find me on GitHub, X (@travisennis), and LinkedIn.
Feedback and collaboration on acai and cake are always welcome — and I'm always happy to talk about AI and software engineering, whatever brings you here.