Field-tested templates and architecture patterns for running an enterprise AI enablement program — from prompt governance and workshop delivery to Power Automate pipelines, Copilot Studio agents, and ROI measurement.
Built and refined while standing up an AI literacy and adoption program from scratch at a global manufacturing company: 29 courses across 11 sites, 1,000+ employees reached, attendance conversion lifted from 53% to 72%.
Most AI enablement efforts stall in the same places:
- A wiki page of "cool prompts" no one can find or govern
- Workshops that demo features instead of building real workflow skills
- Power Automate flows that work in a demo but break under DLP, dedup, or SLA reality
- Copilot Studio agents with no scope, no out-of-scope handling, and no test plan
- ROI conversations with leadership that have no baseline and no story
This toolkit is the operating system I wished existed when I started — concrete frameworks, schemas, expressions, and rituals that take an enablement program from "we bought licenses" to "people use these tools every day, responsibly, with measurable value."
| Audience | What you'll get out of it |
|---|---|
| IT Project Managers / Enablement leads | A reusable program blueprint: governance, measurement, workshop cadence, adoption playbooks |
| Power Platform / M365 makers | Battle-tested Power Automate patterns and Copilot Studio agent designs that survive enterprise DLP |
| HR / L&D / Change managers | Workshop scripts, department playbook templates, and the human-side language that drives adoption |
| Executives / sponsors | An honest measurement framework, proxy-metric models, and an ROI template that holds up to scrutiny |
| Recruiters / hiring managers | A working portfolio of how I think about enterprise enablement — the artifacts, not just the résumé bullets |
| Folder | What it is | Stack / Domain |
|---|---|---|
| prompt-library/ | Governed prompt-library framework: metadata schema, ROI tags, responsible-AI tags, submission/retirement lifecycle, usage measurement | Microsoft 365 Copilot, SharePoint, prompt engineering |
| workshop-facilitation/ | Session planning templates, 90-minute hands-on lab flow, lab exercise schema, multi-site rollout campaign model, pitfall guide | Facilitation, L&D, change management |
| adoption-playbooks/ | Co-created department playbook template (context, quick wins, guardrails, 90-day milestones, escalation paths) and the workshop that produces it | Adoption strategy, responsible AI, change enablement |
| measurement-frameworks/ | Three-tier metrics model (activity / adoption / value), proxy-metric methodology, ROI estimate template, executive dashboard | Program management, analytics, exec reporting |
| copilot-agents/ | Reusable Copilot Studio / Agent Builder patterns: PMO agent, email-intake agent, template-conversion agent, system prompts, topic design, test checklist | Copilot Studio, Agent Builder, prompt design |
| automation-patterns/ | Power Automate architecture patterns: email-to-SharePoint intake, SLA tracking, Forms→List→Excel pipelines, scheduled digests; with key expressions and DLP workarounds | Power Automate, SharePoint, Microsoft 365 |
Each folder has its own README with usage context, schemas/expressions, implementation notes, and testing guidance.
- Microsoft 365: Copilot, SharePoint Online, Outlook (shared mailbox), Forms, Excel
- Power Platform: Power Automate (Standard connectors, Solutions, connection references), Copilot Studio, Agent Builder
- Governance & DLP: enterprise DLP-aware design, lookup-list-driven configuration, registration workflow
- Data / reporting: SharePoint lists as systems of record, Excel/Power BI for analytics, proxy-metric ROI models
- Prompt engineering: structured prompt schema (Context / Task / Constraints / Format), responsible-AI tagging, scoped agent system prompts
- Move people from awareness to fluency. Workshops are designed to send participants back to their desks with a task they'll try AI on tomorrow, not just a demo memory.
- Make adoption legible to leadership. The measurement framework gives program managers a credible story even when clean baselines don't exist — which is most of the time.
- Govern prompts and agents like managed assets. Every prompt has an ID, owner, ROI tag, responsible-AI tag, and a retirement path. Every agent has explicit scope and out-of-scope behavior.
- Ship automations that survive enterprise reality. Patterns assume Standard-connector-only DLP, IT registration, shared-mailbox permission quirks, and the dedup / SLA edge cases that show up on day three.
- Reduce the trust gap. Guardrails, escalation paths, and the "will AI replace me?" conversation are built into the templates — not bolted on as a compliance afterthought.
This toolkit treats responsible AI as a workflow concern, not a policy binder:
- Prompt entries carry data-sensitivity, PII, and review-required tags as first-class metadata
- Department playbooks force teams to write their own "never put this in AI" list during a co-creation workshop — ownership beats compliance theater
- Agent patterns require explicit out-of-scope responses so agents redirect instead of hallucinating
- Workshop flow allocates time to the human side (job-impact concerns, skill embarrassment, trust) before the technical content
- Measurement explicitly names assumptions, confidence levels, and intangible benefits so leadership conversations stay honest
- Browse the folder relevant to your role — each README is self-contained
- Adapt, don't copy — these are frameworks, not finished products; your culture, risk tolerance, and AI maturity will shape the implementation
- Start with measurement — define how you'll prove value before building the program; measurement-frameworks/ is the place to start
- Pair frameworks with each folder — e.g., a department playbook (adoption-playbooks) plus a hands-on lab (workshop-facilitation) plus 5 governed prompts (prompt-library) is a complete adoption motion for one team
If you're evaluating me for an IT Project Manager, IT User Enablement, AI Enablement, or Power Platform / Copilot program role, this repo is intended to demonstrate:
- Program design under enterprise constraints — DLP-aware automation, governed prompt library, registered flows, scoped agents
- End-to-end enablement thinking — from awareness session to department playbook to measured ROI, with the change-management glue between them
- Hands-on Microsoft 365 / Power Platform craft — actual expressions, schemas, topic designs, and architecture diagrams, not just slideware
- Responsible AI as a default — embedded into every artifact (prompt, agent, playbook, workshop)
- Measurement honesty — proxy metrics, named assumptions, and an ROI template that doesn't oversell
- HR-to-IT translation — 11 years in HR/Benefits before IT means the workforce side of adoption isn't theoretical for me
Happy to walk through any folder live and talk through trade-offs and what I'd do differently next time.
I'm Wes Shelton — an IT Project Manager / IT User Enablement professional focused on AI workflow, automation, Microsoft 365 / Copilot, Power Automate, and enterprise adoption programs. I designed and delivered the AI literacy and adoption program these templates come from, and currently work on AI-driven workflow improvement and enablement at scale.
Before moving into IT, I spent 11 years in HR (Benefits at Rocket Companies, Wayne County, Blue Cross Blue Shield), which informs how I approach the trust, skill-gap, and workflow-integration side of adoption.
All templates here are reconstructed and generalized from real program work. No client-, employer-, or employee-specific data is included. Company names, employee data, and proprietary configuration have been replaced with placeholders ([Company], yourdomain.com, sample IDs). Adapt before use.
MIT License — use, adapt, and share freely. Attribution appreciated but not required.