WORKING GUIDE
How to run an AI transformation in 28 weeks
A practical route from a measured process baseline to an AI agent in production — with owners, gates and a week-by-week plan.
The transformation map
The route itself is open. Four phases, 12 steps; the timeline below is calibrated for an organisation of 160 users.
| Phase | Steps | Timeline (160 users) | Phase result |
|---|---|---|---|
| A. Decisions and preparation | 1–3 | Weeks 1–4 | Management order: 4 processes, 4 agents, 4 C-level owners; target ROI metrics approved |
| B. Digital foundation | 4–6 | Weeks 5–16 | 4 processes run in d8n in production; the corporate knowledge base is built |
| C. Agent launch | 7–9 | Weeks 17–28 | 4 agents configured, piloted and rolled out to all users |
| D. Operations and ROI | 10–12 | Ongoing | Weekly working groups run; ROI is measured and defended quarterly |
The full cycle takes 6–8 months from the management order to measurable ROI across all four agents.
Download the file
Get the DATM guide and 28-week plan
Enter your email. We will send a one-time link valid for seven days; no attachments.
The 12 steps
Phase A — decisions and preparation
- Management decision and order
- Readiness assessment (discovery)
- Target metrics and the business case
Phase B — digital foundation
- Deploy the standard d8n configuration (foundation)
- Run the processes in production
- Corporate knowledge base and document catalogue
Phase C — agent launch
- Agent configuration and calibration
- Pilot with AI champions
- Full wave-by-wave rollout
Phase D — operations and ROI
- A weekly working group per agent
- ROI measurement and the three-layer metric model
- Quarterly review and scaling
What is inside the file
- A 28-week Excel schedule for owners, dependencies, gates and acceptance.
- Checklists for process readiness, knowledge quality and production launch.
- Quality thresholds for hallucinations, relevance, citation accuracy, adoption and NPS.
- Templates: the order, the weekly working-group protocol and the risk register.
