Enablement as a product: every build makes the next one easier
AISE doesn’t only ship agents. Each build leaves behind a standard, a training module, a reusable skill or a capability another team can run, so the next build, ours or anyone’s, starts further ahead.
Why enablement matters
A three-person team can only build so much by hand. What scales is making each build repeatable: a written standard for the platform, a skill that can be lifted into the next agent, a guide a new engineer can follow on day one. The director role is “AI Enablement & Knowledge,” not just engineering, for exactly this reason. The team’s output is agents plus the ability for more people to build them well.
Standards written from real builds
The standards live in the team’s core repository, and nearly every one exists because a build hit something the documentation didn’t cover. HR Studio found that uploading a skill as a .zip silently strips its instructions, which became the new-experience build guide. AutoBid’s promotion found that a managed import points retrieval tools back at the environment they came from, which is now part of the deployment pipeline. The Data Assistant was the team’s first Azure Function, and its lessons log became the Azure Functions standard the week it shipped. There are 20 standards today, covering Copilot Studio, Claude Managed Agents, Azure Functions, Rovo, solution export and import, and how the team runs Claude Code across many repositories.
“When something bites, write it down where the next person will look.”
— AI Solutions Engineering
Training that gets people shipping
New engineers and interns follow a ten-day onboarding curriculum with daily follow-alongs, plus step-by-step guides: clone a repo, install the toolchain, create an agent, edit one, replatform one. The summer interns shipped flagship work, including SAM and the IT Support agents, within weeks of starting.
A library of skills anyone can reuse
The Skills Library is a published catalog of 103 agent skills: 42 built across the AISE portfolio and 61 curated from Microsoft. Each is copied word for word from the agent that uses it, so what’s in the library is what’s proven in production, ready to drop into the next agent.
Capabilities other teams can run
Some of what the team builds isn’t an agent at all. Knowledge Harvesting, which turns a shared mailbox into documentation, is packaged as a set of Claude Code skills a teammate installs and runs on their own machine. The team also uses its own skills for delivery work: scaffolding a new agent repository with its governance documents, planning sprints, and drafting executive summaries. The practices it recommends to others are the ones it runs every day.
What it adds up to
Each new build starts from the last one’s lessons instead of rediscovering them. It is why a small team can run more than 30 agents across five platforms, and why the next team at MillerKnoll that wants to build doesn’t have to start from zero.