A working knowledge base on two intertwined ideas reshaping how organizations get value from AI:
- AI enablement: the discipline of helping teams and organizations actually adopt AI, beyond buying licenses and hoping.
- Forward deployment: the practice of embedding engineers directly with the teams or customers who have the problem, building with them rather than for them, and feeding what's learned back into the platform.
These two ideas answer the same question from different directions: why does so much AI investment fail to change how work gets done, and what does it take to close that gap?
Most organizations are discovering the same uncomfortable truth: the model is not the bottleneck. Access to frontier AI is nearly universal; transformation is rare. The gap is organizational (workflow, trust, skills, incentives, and the "last mile" of integration into real work). This repo collects frameworks, playbooks, and field notes for closing that gap.
This is a knowledge project, not a software project. Everything here is prose, templates, and models meant to be read, argued with, forked, and improved.
| Section | What's in it |
|---|---|
enablement/ |
What AI enablement is, a maturity model, an adoption playbook, failure modes, how to measure impact, and sprint zero: setting up agent-ready projects |
forward-deployment/ |
The forward-deployed engineering (FDE) model: its history, the engagement lifecycle, team shapes, and how to run it inside your own org |
playbooks/ |
Ready-to-use templates: discovery question bank, engagement charter, pilot scorecard, handoff checklist, sprint zero checklist |
GLOSSARY.md |
Shared vocabulary, because half of the confusion in this space is terminological |
READING.md |
Curated sources and further reading |
- If you're asking "how do I get my org to actually use AI?", start with enablement/00-overview.md
- If you're asking "what is a forward-deployed engineer and should we have them?", start with forward-deployment/00-overview.md
- If you're about to kick off an AI project with a team, grab the templates in playbooks/
- If you're asking "how should a project be set up so AI agents can actually work in it?", start with enablement/05-sprint-zero.md, then see the concrete reference setup
These are the claims the rest of the repo elaborates and defends:
- Adoption is a supply problem disguised as a demand problem. People want leverage; what's scarce is someone who translates a general-purpose capability into their workflow.
- The unit of transformation is the workflow, not the tool. Deploying a chatbot changes nothing; redesigning how a claim gets processed changes everything.
- Embedded beats centralized, but only with a feedback loop. A center of excellence that never ships alongside the teams it serves becomes a slide factory. Forward deployment is the antidote, if learnings flow back into shared platforms.
- The engagement must be designed to end. Success for an enablement or FDE engagement is the embedded team becoming unnecessary. Dependency is failure with good optics.
- Measurement keeps you honest. Adoption theater (seats provisioned, prompts sent) is easy. Outcome measurement (cycle time, quality, rework) is what separates transformation from tourism.
This is a living document set. Opinions are held with confidence and revised with evidence. Content is licensed CC BY 4.0.