The 90-day rollout plan

Three phases, one workflow at a time · Last updated 2026-08-12 · Ousios LLC

Most AI rollouts fail the same way: too many tools, too fast, with no agreed measure of whether any of it worked. This plan does the opposite. One workflow, proven in the first 30 days, or you kill it and pick another.

The structure below is fixed. What changes per team is who does what — which archetype champions adoption, which one owns documentation, who audits risk. Composa computes that from your team's actual composition. You can read the whole plan here first and decide whether it is worth mapping your team at all.

Days 1–30 — Prove it

Goal: one workflow, one win, zero risk.

Pick the single most time-wasting recurring task. Your team's dominant strategy is the lens for choosing.
Run it in a sandbox with sample data first — no live stakes.
A champion drives adoption; everyone tries it twice. Composa names who that should be — typically the Pioneer or the Catalyst, because adoption is a social act before it is a technical one.
Measure hours saved. If it doesn't save real time, kill it and pick another.

Days 31–60 — Systematize

Goal: turn the win into a repeatable process.

One person owns turning the pilot into a documented, default workflow. Composa names the archetype best suited to it.
Add the second workflow — ideally one that covers a team blind spot rather than doubling down on an existing strength.
Set the quality bar: who reviews AI output before it ships?
Train every archetype on their personal AI move.

Days 61–90 — Scale & protect

Goal: expand where it works, guard where it doesn't.

Roll the two proven workflows to the whole team.
Someone owns risk explicitly: data handling, output quality, over-reliance. If your team has a Guardian, this is their phase.
Book the 90-day re-assessment. Teams drift, and the strategy should drift with them.
Set the next quarter's single AI goal, based on what actually moved the needle.

Why three phases and not a checklist

Each phase has a different failure mode, and they need different people. Days 1–30 fail on enthusiasm — a team tries six tools and proves nothing. Days 31–60 fail on drift — the pilot worked but nobody wrote it down, so it dies when its champion goes on holiday. Days 61–90 fail on over-reach — the team scales a workflow past the point where anyone is checking the output.

Matching the phase to the person who is temperamentally suited to it is the entire idea. A team of Pioneers will nail the first 30 days and quietly skip the second 30.

What this plan is not. It is not a guarantee, and it is not a benchmark. Whether it works depends on implementation quality and team adoption — the two things no plan can know in advance. The published list of what Composa cannot see, cannot know, and cannot prove yet is at What Composa won't tell you.

Map your team and get the personalized version →

If you would rather commit to less first, the 2-week pilot is the same discipline at a smaller scale: you define pass/fail before you start.