How do we scale AI past a few keen people?
November 2026
The short answer
You scale AI by making what works for a few people work the same way for everyone: shared skills, connections to your real systems, an owner for each piece, and limits on what it costs. More licences on their own do not do it.
Why it stalls
The usual picture: a handful of people use AI every day and are flying, everyone else tried it once and went back to how they worked before. What the keen people know lives in their own prompts and chats, so nobody else benefits.
What scaling actually takes
- A shared skills library. Take what your best users do and turn it into skills everyone uses. That is how one person’s good habit becomes the team’s way of working.
- Connections to your systems. AI that can read your jobs, customers and numbers is useful every day. AI that has to be fed copy-and-paste is not.
- An owner for every skill, agent and connection, who keeps it current when the process or the tool changes.
- Cost controls. Copilot, Notion, monday.com and HubSpot already bill AI by use, and Atlassian starts on 3 December. Set limits per person or per feature, and check usage monthly.
- Governance people actually follow. Clear rules on data, approvals and which tools are allowed, taught, not just written down.
- Someone watching the tools. They change every month. Somebody has to read the release notes and decide what it means for you.
Measure the work, not the logins
Licence dashboards tell you who opened the app. They do not tell you whether quotes go out faster or month-end closes sooner. Pick the measure that matters for each task before you scale it, and check it after.
Where we come in
This is the build and maintain side of what we do: writing the skills with your people, connecting the systems, setting the guardrails and cost limits, and keeping it all current as the tools change. Our monthly tool updates are part of that.