The companies getting value from GenAI are not the ones with the grandest strategy decks. They are the ones that shipped one useful thing, learned from it, and shipped the next. A 90-day roadmap is how you join them.
Days 1–30 — Find the wedge
Pick one workflow where the value is obvious and the risk is contained: a support assistant, a document summariser, an internal search. Define what success looks like as a number before you build anything.
Days 31–60 — Build the thin slice
Ship the smallest version that delivers real value to a real user, with the safeguards that matter:
- Wire up the workflow end to end for a single team.
- Add evaluation and guardrails from day one, not as a later phase.
- Put it in front of real users and measure against the success number.
Days 61–90 — Harden and expand
Use what you learned to make it reliable, then widen the audience. By day 90 you should have one workflow in production, a metric that moved, and a repeatable playbook for the next one.
GenAI strategy is not a document. It is the second workflow you ship, faster than the first.
Why initiatives stall
- Trying to boil the ocean instead of scoping to one workflow.
- Treating evaluation and guardrails as a later phase, so nothing is safe enough to ship.
- Optimising the demo instead of the production path, so the pilot never graduates.
A quarter is long enough to ship something real and short enough to force focus. Use the pressure: pick the wedge, ship the slice, harden it, and let the playbook carry you into the next workflow.
