Only about 36% of US workers with Copilot at work actively used it in a January 2026 survey of more than 150,000 people. Unused seats are money you've already spent. Recon Analytics
What we measure
Proc scores how each person works with the AI, from the first request to the final check. Then it coaches them on their own prompts, starting with the change that will help most.
Says what they want, in their own words.
Gives the AI what it needs to know about the situation.
Sets the audience, length, tone and format.
Tells the AI exactly what to change when the first answer falls short.
Checks what the AI assumed and whether the answer fits.
Built for your team's work
Before the pilot starts, we write the tasks for your team's department. Each one is a realistic job task with sample data, worked through with an AI assistant built into the page.
Review this week's 15 expense reports and flag the charges that need a second look.
Turn this quarter's pipeline export into a forecast summary for your VP.
Pull the lessons from a campaign's results and draft next month's plan.
Sort a messy thread of requests into priorities and a plan for the week.
Everything happens inside Proc, with sample data. Proc never connects to your systems or sees company data.
How a pilot runs
A test built around your team's work.
One new skill a week: a short lesson, then practice tasks with personal coaching.
New tasks of the same difficulty, so gains reflect real skill.
Where the team started, where it ended, and what to work on next.
Five weeks, from baseline to re-test and a team report.