Agentic systems people can trust
I build with LangGraph, LangChain, Pydantic AI, and MCP servers. To that I add tool calling, structured outputs, human-in-the-loop controls, evaluations, and guardrails.
Gary Butler
Software engineer. AI builder. Always learning.
More than 30 years developing software and now bringing that experience to hands-on work with AI.
30+Years building software
17Years leading teams
4Projects in the open
01What I do
A builder's depth. A leader's perspective.
My work spans applied AI, enterprise systems, technical leadership, and the projects that connect them.
View full experience →RAG that cites its work
Vector databases and retrieval pipelines built around enterprise data, so every answer traces back to its source.
Teams that ship and support
Seventeen years leading developers and analysts through enterprise delivery, modernization, and production support.
Legacy systems, brought forward
Agentic AI coding tools pointed at aging applications, rebuilding them rapidly on a modern stack.
Rapid prototypes and proofs of concept
Built quickly with agentic AI, so a rough idea becomes something you can actually put in front of people.
02AI Projects
Small builds. Useful evidence.
Each one exists to answer a question I could not answer by arguing about it.
View all projects →HazardAI
Can a vision model turn job-site photos into a useful first draft of a job hazard analysis without taking the final decision away from a person?
Vision AIReact NativeHuman review
View project ↗Signal Desk
Can an agent turn a noisy set of sources into a short brief without hiding where its claims came from?
AgentsResearchCitations
View project ↗Brief Builder
Can a model help shape an early product idea without inventing certainty?
Structured outputProduct discoveryLLMs
View project ↗Model Lens
Can model selection be explained by the job, cost, latency, and failure risk instead of a leaderboard?
EvaluationModel choiceCost
View project ↗03Writing
Notes for navigating the AI shift.
Practical ideas for people designing, leading, and making sense of AI-powered products.
View all writing →A second set of eyes, not the final word
What I learned building a vision-assisted job hazard analysis that keeps people responsible for every final decision.
Read essay ↗The best AI agents know when to stop
Autonomy gets the attention. Good judgment—and a clear boundary—earns the trust.
Read essay ↗Prototype the question, not the platform
A faster way to learn whether an AI idea is valuable before building the machinery around it.
Read essay ↗04Experience
Software has to keep working after the demo.
I have spent more than 30 years building software and 17 years leading development and analyst teams. That work taught me to care about permissions, failure paths, handoffs, and the people who support the system later.
View my experience →Built around real systems
My background is in enterprise applications, integrations, modernization, and production support. I build on the systems a team already runs rather than ripping them out and starting over.
AI with a visible boundary
My recent work covers LLM applications, retrieval, tool-using agents, structured outputs, evaluations, and human approval. I use those tools when they fit the job.
Beyond the résumé
The work matters.
So does how I think.
My essays and project notes show the questions behind the systems: what AI should do, where it should stop, and what useful evidence looks like.
Read my writing ↗