# AI Implementation Roadmap
A sequenced, phase-by-phase plan for organizations adopting AI. Not theory. Clear deliverables and success criteria at each stage.
## Phase 1: Foundation (Weeks 1-4)
**Goal:** Establish baseline and make one thing work.
- **Assess current state.** What tools are people already using? Where is [[Shadow AI]] happening? What data flows exist?
- **Define [[AI Usage Policy]].** What's allowed, what's not, what data can touch AI systems. Get legal sign-off.
- **Pick one use case.** Choose the highest-impact, lowest-risk workflow. Code review, documentation, customer support triage; something concrete.
- **Select tools.** Standardize on 1-2 AI platforms. Avoid tool sprawl from day one.
- **Set up approved accounts.** Enterprise licenses, SSO, audit logging. No personal accounts for work data.
**Deliverables:** AI usage policy document, approved tool list, one pilot use case selected, accounts provisioned.
**Success criteria:** Policy published, tools accessible, pilot team identified.
## Phase 2: Pilot (Months 2-3)
**Goal:** Prove value with one team before scaling.
- **Run a pilot with one team.** Small, motivated, technically capable. Let them break things.
- **Establish team context.** Shared CLAUDE.md or equivalent with coding conventions, processes, priorities. See [[Team Context Management (TCM)]].
- **Measure results.** Time saved, quality changes, adoption rate, user satisfaction. Quantify everything.
- **Iterate.** What worked? What didn't? What context was missing? Refine the approach before expanding.
**Deliverables:** Pilot results report, refined context files, documented lessons learned.
**Success criteria:** Measurable improvement in at least one metric, team willing to continue.
## Phase 3: Expansion (Months 4-6)
**Goal:** Scale what works to additional teams.
- **Expand to additional teams.** Use pilot learnings to onboard 2-3 more teams.
- **Build shared skill library.** Codify repeatable AI workflows as reusable skills. See [[AI Agent Skills]].
- **Deploy team-level agents.** Each team gets agents configured for their domain, with appropriate context and permissions.
- **Cross-team knowledge sharing.** What one team learns should benefit all teams.
**Deliverables:** Skill library with 10+ reusable skills, 3+ teams actively using AI, cross-team sharing process.
**Success criteria:** Consistent adoption across teams, skill reuse happening organically.
## Phase 4: Enterprise (Months 7-12)
**Goal:** AI as organizational infrastructure.
- **[[Enterprise Context Management (ECM)]].** Organization-wide context hierarchy: org level, division level, team level, individual level.
- **Governance framework.** See [[AI Governance]]. Audit trails, compliance checks, regular policy reviews.
- **Org-wide rollout.** All teams have access, training, and support.
- **Continuous improvement.** Regular review cycles, skill library growth, context refinement.
**Deliverables:** Enterprise context management system, governance framework, org-wide training program, quarterly review process.
**Success criteria:** AI integrated into standard workflows across the organization, measurable ROI.
## Scaling Notes
**Solopreneur:** Skip to Phase 1 essentials. You need a usage policy (even if it's just for yourself), one tool, and one use case. Context management is your competitive edge from day one. See [[Levels of AI use]].
**Small team (2-10):** Phases 1-2 compressed into 4-6 weeks. Phase 3 is your steady state. Phase 4 is optional.
**Enterprise (50+):** Full four phases. Don't skip Phase 2. The pilot protects you from scaling mistakes. See [[Enterprise AI Deployment]] and [[AI Transformation Playbook]].
## References
## Related
- [[AI Usage Policy]]
- [[Enterprise AI Deployment]]
- [[AI Transformation Playbook]]
- [[Levels of AI use]]
- [[Team Context Management (TCM)]]
- [[Enterprise Context Management (ECM)]]
- [[AI Agent Skills]]
- [[Shadow AI]]