The problem
Copilot content online splits into two useless camps. Camp one is the press release: everything is magical, adoption is effortless, your org will save 14 hours per person per week. Camp two is the hot take: it's all hype, nothing works, cancel the licenses.
Reality is in the middle, and it's specific. Teams meeting recap is genuinely one of the best software features shipped this decade. Excel Copilot will embarrass you in front of your CFO if you don't know its constraints. Declarative agents can deflect half your IT tickets — or confidently serve your staff a policy that was retired in 2023, depending entirely on whether anyone audited the SharePoint site first.
Knowing which is which, surface by surface, is the entire game. That's what we publish.
What we promise
- Scores you can act on. Every Truth scorecard says who should use the tool, who should skip it, and the workarounds for the gaps.
- Currency. The agent ecosystem changes monthly. We track manifest schema versions, billing model changes, and feature GAs — and date-stamp everything.
- Mechanics, not magic. Prompts come with the why. Guides explain how grounding and the semantic index actually behave, so you can reason about failures instead of memorizing templates.
- Installable, not aspirational. Agent blueprints ship with manifests, instructions, deployment steps, and governance notes — tested designs, not conference demos.
Who's behind it
Practitioners who deploy this stack in real organizations — with the scar tissue, failed pilots, and successful rollouts to prove it. Independent: we are not affiliated with Microsoft, and nobody pays us to round scores up.
What's coming
A growing blueprint library, adoption measurement templates, deep-dives on Copilot Studio's autonomous and computer-use agents, and — eventually — a pro tier with full rollout kits. The field manual core stays free.