Anonymized for now, real throughout. Names and logos appear as clients clear them.
The situation. An internal survey confirmed it: almost everyone using personal ChatGPT accounts. Sequences, structures and clinical data are the core IP, one casual prompt away from a leak, and a new-market expansion was raising the stakes.
What we did. Two tracks. A standardized, properly configured platform with a usage policy people actually read, moving fast. Parallel advisory on the sensitive-data infrastructure questions, moving carefully.
What changed. Team-wide fluency and working prototypes for the priority workflows.
The situation. ChatGPT Enterprise active, teams experimenting, no systematic approach, and no dedicated digital leadership in-house to coordinate one.
What we did. Fluency sprint, anonymous readiness survey with a management readout, then a live MVP hackathon that built prototypes with the team in the room.
What changed. From exploration to systematic use, with a prioritized use-case backlog.
The situation. Organizations across regulated industries needing baseline AI fluency that sticks past the workshop day.
What we did. Role-specific workshops built from pre-workshop interviews, so every session runs on the participants' real tasks.
What changed. The pushback is always the same at the start. The results, when people actually experience the tools, tend to speak for themselves.
The situation. Copilot in every Office app, and a fraction of it used: some chat on the web, almost nothing in Excel, PowerPoint, Teams or Outlook, where the actual work lives.
What we did. Five one-hour sessions, each ending with the tool doing real work: prompt technique and hallucination handling, a first agent built live in minutes, Copilot in Excel and PowerPoint with honest coverage of its limits, and the meetings-to-decisions workflow in Teams and Outlook.
What changed. The licence the company was already paying for started earning its keep. Everyday tasks moved into Copilot, with frameworks the team now runs weekly.
The situation. Leadership suspected AI use was happening off the books but had no picture of who, where, or with which tools. An official AI policy existed; almost nobody knew what it said.
What we did. An anonymous five-minute readiness survey: 107 responses across quality, finance, commercial, regulatory and HR, mapping usage levels, shadow tools, policy awareness and hidden champions. Then a management readout that turned the data into a plan: clarify the existing policy, focus the tool strategy, a three-tier training programme, and champions activated into a community.
What changed. The site went from guessing to knowing: a 58% adoption baseline, named champions, and a governance fix aimed at the real blocker: unclear rules, ahead of missing skills.
The situation. Commercial teams at two global health companies wanted their people using AI on real tasks by the next morning, without another slideshow about the future of work.
What we did. The AI in Action format, prepared through 30-minute pre-session calls so the use cases on screen are the team's own: how the models work and where they fail, prompt technique with live exercises, reusable prompts for the tasks that repeat, and a first custom assistant built as a live demo. Live polling opens and closes the session, so the team sees its own confidence shift.
What changed. Participants leave with prompts and an assistant they built themselves, scoped to their roles. The format has since run across pharma, hearing care and professional services teams in multiple countries.