AI Prompt Hub & Copilot Training

BARBRI rolled out Microsoft Copilot and Claude to staff, and the same thing happened that happens everywhere: a handful of people got brilliant at it, everyone else typed one vague sentence, got a mediocre answer, and quietly went back to doing things by hand.

I designed and built two connected platforms to close that gap. The AI Prompt Hub is a living, department-organised library of tested prompts. The Copilot Foundational Training programme is eight interactive sessions that teach staff how to think in prompts. Each one links to the other, so learning and doing never sit in separate tabs.

Good prompts existed at BARBRI. They lived in private notepads, Slack threads and one tutor’s legendary Word doc. A tutor who had perfected a marking-feedback prompt had no way to hand it to thirty colleagues, and a new starter in Finance had no way to find it.

Training was the other half. Generic “intro to AI” videos don’t stick because they aren’t about your job. Staff needed to see Copilot working on their documents, their emails, their marking, and they needed somewhere to put that skill to work the same afternoon.

The design bet was personalisation over a generic library. The hub opens by asking who you are and what you work on, and from then on it leads with your department: tutors see marking and feedback prompts, HR sees policy drafting, Finance sees reconciliation helpers. A shared library that feels personal gets used; a generic one gets bookmarked and forgotten.

The second bet was that quality needs owners. Every department has a champion who reviews contributed prompts before they go live, seeded examples are labelled as placeholders until a champion replaces them with tested ones, and searches that return nothing are logged as prompt requests so the library grows where demand actually is.

The hub makes prompts usable, not just findable. Every prompt page has fill-in-the-details fields: type your student’s name and word count, and the prompt rewrites itself live, ready to copy. Ratings, copy counts and a discussion thread sit underneath, so the best version of a prompt keeps improving in public.

The hub turned prompt knowledge from private property into shared infrastructure. The best marking-feedback prompt at BARBRI is no longer in someone’s notepad; it’s rated 4.6, copied by 31 colleagues, and improved twice in its own discussion thread. And because training hands off directly into the library, staff leave a session with somewhere to go.