Process-level agentic transformation. With evidence.
Working alongside your team to build and launch AI agents, with pricing tied to measurable business outcomes.
01 / What we take on
These are the ones we see most often. If your process looks different, we would still like to hear about it.
High volume, low judgement, and the only way to do more of it is to hire more people.
Rebuilt so agents do the reading and the fetching, and your team makes the calls that actually need a person.
It works on the examples it was built against. Nobody will put real traffic through it.
Taken into production against your real cases, or a straight answer on why it should not be.
It is live, it is useful, and no one can tell you how often it is right or what it does when it is wrong.
Scored against a written definition of correct, so trust becomes a number you can take to a board.
They worked on launch day. Since then the data moved, the prompts changed, and the costs are a line item nobody can explain.
Quality and cost visible in production, with a gate that stops a bad change before it ships.
02 / Selected work
What we rebuilt, what it changed, and the numbers to show it.
A multi-location Mexican fast-casual chain in Southern California
A financial services firm's AI platform team
03 / Team
Partner · Los Angeles
Senior engineer and tech lead at Amazon, with over a decade across full-stack, cloud, MLOps, and DevOps. He has led technical teams and mentored hundreds of developers, from bootcamp graduates to senior engineers. Alongside that he consults as a machine learning engineer, designing and building agentic systems, RAG pipelines, and the evaluation systems that tell you whether they work.
Partner · Montréal
A product and technology leader with a decade at Expedia Group. Seven years engineering the systems behind hotel pricing and recommendations, then leading the AI evaluation and quality frameworks that put AI-driven travel in front of millions of travellers. He has taken AI from prototype to production at consumer scale, and set the quality bar it had to clear first.
04 / Services
Every step ends with something you keep. Begin at whichever one fits where you are today.
05 / Blog
What we have learned doing this work, explained plainly. Take what is useful.
GenAI output is probabilistic, so the old question, did it match the label does not work. Why AI products fail silently, and why spot-checking cannot catch it.
A stack, seven product archetypes, and the one question that decides most of your eval design: is there a right answer to compare against?
Four principles, how to build an LLM-as-a-judge forwards rather than backwards, the five metrics to start with, and why robustness comes from stacking imperfect layers.
Start with a conversation
Free, and useful whether or not we work together. Bring the process that keeps costing more than it should, and we will tell you honestly what we would do about it.
No deck. No discovery workshop. Just your process.