A pilot that worked — and couldn't scale itself.
A fintech ran a successful AI-native pilot on one team. The results were real, but the playbook lived in a few people's heads and the tooling was hand-rolled. The pilot proved the model; nothing existed to scale it.
They needed two things in sequence: the operating model installed across the org, and the infrastructure to run it at that size.
Advisory installed the model. Studio built the system.
First, Advisory turned the pilot's tacit playbook into installed conventions, reviews, and evals across every team. Then Studio built the shared infrastructure — pipelines, guardrails, and a scoreboard — that let the whole org run the model without a champion in every room. Same buyer, two verbs: one taught, one built.
- Advisory: the pilot's playbook, written down and installed org-wide
- Studio: shared pipelines and guardrails the whole org runs on
- A scoreboard that makes the rollout visible team by team
- A handover where the org owns both the model and the system
"By the second quarter, AI-native wasn't a team — it was just how the company shipped."
Teach first, then build to scale.
The sequence matters. Build before the model is installed and you scale a habit nobody has yet; install without building and the model stalls at the size one champion can hold. Advisory then Studio is what got it to org-wide.