Victor Gless-Krumhorn

I founded AI-Native Agency in 2026 after cofounding Jaikin earlier the same year, and two years as an Analyst at UBS. I studied at Université Paris Dauphine — PSL, where I trained on the quantitative and decision-sciences side of finance.

My thesis is simple: an AI-native agency is not a consulting firm with an LLM wrapper. It is an operatorthat ships and runs the workflow every week, with KPIs published against baseline and controls a risk officer can audit. Engagements are scoped, fixed-priced, and delivered end-to-end by our team — never referred to a third party.

What we build, we operate. The agency's own stack — from client intake to RAG pipelines, prompt versioning, evaluation harnesses and weekly performance reviews — comes out of our keyboards. Daily stack: Next.js, TypeScript, Python, FastAPI, PostgreSQL, Claude/OpenAI multi-LLM, Docker, Vercel.

The two years inside UBS shaped how I think about delivery. In a regulated environment, the wrong process is more expensive than no process at all. That filter — would this survive a risk officer's review?— is now embedded in every engagement we ship: approved sources, versioned prompts, audit-ready logs, reviewer queues for low-confidence outputs, and quarterly attestations on request.

If you want to talk about a specific workflow you want to attack first, book a discovery call. I read every brief myself.