AI transformation / Planning worksheet
AI adoption: a 90 day roadmap
Your team is trying AI. What would it take to make it a useful part of the work? A suggested 90 day plan, from choosing a pilot to deciding what deserves to expand.
One9Six / Working notes
Define an operating model for AI
Start with one repeatable task and a baseline measured without AI. Record time spent on drafting, review and correction separately. Speed only becomes useful when the approved output meets the same quality standard. NIST’s framework offers a governance structure, not a certification that a workflow is safe.
01Baseline
02Controlled pilot
03Evaluation
04Approved use
| Focus | Record | Practical check |
|---|---|---|
| Input boundary | Approved source set and permitted data categories | Confirm access, retention and reuse terms before supplying confidential material |
| Evaluation | Representative briefs, factual errors and review time | Use the same test set when comparing models or prompts |
| Release control | Named approver and retained final version | Log model, prompt version, sources and reviewer decision |
Apply this to your next project
- Treat retrieved documents as evidence, not as instructions to the model.
- Define a stop condition for unsupported claims or exposure of restricted data.
- Expand only after a documented review of quality, cost and responsibility.
Make it practical
A useful place
to start.
Use this guide with the colleagues involved in the work. Note what you know, what needs checking and who will take the next step.
Download PDF- Days 1 to 10 / Map the work. Choose recurring tasks such as briefing, research synthesis, content adaptation or reporting. Record inputs, handovers, review time and recurring problems before selecting a tool.
- Record how the work is done today. Measure total time per completed task, including corrections and approvals. Agree what good output looks like and record current quality against the same criteria you will use in the pilot.
- Choose one pilot. Score candidate workflows for business value, frequency, available source material and consequences of error. Start with a contained internal task whose output a qualified colleague can verify.
- Agree the boundaries. Name the business owner, reviewer and technical contact. Define approved tools, permitted information and material that must stay outside them. Verify settings and organisational requirements before using company or client data.
- Days 11 to 30 / Design the workflow. Document the trigger, approved source pack, instructions, expected output and review gate. Specify what happens when information is missing, conflicting or outside the agreed task.
- Build the test set. Use representative examples and difficult cases. Compare outputs with the current process for factual accuracy, completeness, brand voice and usability. Keep a record of failures and the changes made in response.
- Days 31 to 60 / Run a supervised pilot. Train a small group using real examples and an agreed review checklist. Keep publication and external communication subject to named human approval.
- Measure the complete process. Track time saved after review, correction rates, team adoption and total tool cost. Ask whether the output is genuinely easier to use, rather than simply faster to produce.
- Days 61 to 90 / Decide what scales. Expand only the workflows that meet agreed quality and business criteria. Record the owner, training needs, review frequency and fallback process before involving more teams.
- Create the record of the decision. Summarise the workflow, baseline, pilot evidence, unresolved issues and next investment decision on one page. Revisit the evaluation when tools, source material or responsibilities change.