Sharing Your Agent and Building for Others · Building agents in Copilot

One agent, one colleague, one feedback round.

Share with specific people (link) for soft launch; publish to the organization catalog when ready (usually requires admin approval at MillerKnoll).

Designing for others: starter prompts matter more, scope boundaries must be explicit in instructions, first 60 seconds decide adoption.

Template 1: Team knowledge agent: team SharePoint, FAQs, who owns what.
Template 2: Meeting prep agent: past notes and project docs, open issues, decisions.
Template 3: Document drafting agent: templates, brand voice, past examples.
Governance: confirm data sensitivity and permissions; designate who updates knowledge when content changes; confirm catalog approval process.
Culture: share starter prompt library, one-paragraph brief (what it does / does not do), ask colleagues to break it, bad outputs are curriculum.
After Agent Builder: Copilot Studio for Salesforce updates, scheduled workflows, branching logic, external APIs, master Builder first.
Checklist before team share: specific name, 200+ word description, instructions with scope and uncertainty, current knowledge, starters that show best behavior, 10+ tests including edge cases, one colleague reviewed, permissions confirmed.

Check yourself

Do this in Copilot

Build v1 from one template, test per Lesson 4, share with one colleague and ask them to break it.

Paste this into Copilot Chat and work through it before moving on.

Write a one-paragraph brief I can send with my agent share link: what this agent does, what it does not do, when to use it instead of regular Copilot, and three starter prompts to try first.

Open Copilot →

Share brief

Pre-share checklist: name specific · description detailed · instructions include uncertainty · knowledge current · 10+ tests · one colleague feedback · permissions OK.

Write a one-paragraph brief I can send with my agent share link: what this agent does, what it does not do, when to use it instead of regular Copilot, and three starter prompts to try first.

Did you run this in Copilot? Mark complete when you have tried it.

The longer game

You just learned to build something that works without you in the room, a specialist that handles a repeatable task reliably, for anyone, every time.

Here's the scale of what that capability sits inside. The same research that found only 10% of large companies truly excelling at AI readiness identified exactly why the other 90% stall: not access, not tools, not budget. The bottleneck is whether the people closest to the problem have the autonomy to act on what they see. The distance between insight and action, between "this process is broken" and "I built something that fixes it", is what agents collapse. Every agent you build is a small, permanent transfer of knowledge out of someone's head and into something the whole organization can use. The question now isn't whether you can build one. It's whether you can see the friction in your organization clearly enough to know what to build next, and whether you have the ownership to build it before anyone asks you to.

Core principles

One-paragraph share brief

Building agents in Copilot