Northwind
17 / 30
A start, no engine band 10 to 19
Some pieces are in place and some people are good at it, but nothing keeps it moving when the keen people get busy.
3 points to Almost there.
Your first 3 moves
The lowest lines, with what the next level up looks like.
Hours saved are measured and shown to leadership
Now 0 of 3. The next level up:
There are stories, but no numbers.
Training comes in 3 tracks: starters, daily users and builders
Now 1 of 3. The next level up:
There are 2 levels, or the 3 tracks exist on paper only.
A list of AI use cases, ranked by hours saved
Now 1 of 3. The next level up:
The list is ranked by gut feel or by who asked loudest.
Want someone to run the 90 days with you? We run the plan with your champion: set-up, the first 4 workflows timed, and the hackathon.
Share this result
The link holds only the 10 numbers and the team name. Post it in your team channel and ask each lead for their own number.
https://tools.openkrill.app/tenrung/r/3232121210?team=Northwind
Your 90-day plan
The same 3 months for every team. Your scores decide which steps start from nothing and which only need a check.
Month 1
Set up
Give AI work an owner, rules, a place to learn and a weekly rhythm.
- Week 1KeepCheck the champion still has their hours, and that people know who to ask.
- Week 1KeepReread the 1-page policy, add any tool approved since, and pin it again.
- Week 2TightenPick one rep per department and book a 30-minute meeting every 2 weeks for the next 90 days.
- Week 2TightenOpen one shared library for recorded sessions and how-to notes, and give the champion the job of checking it every month.
- Week 3StartPut a weekly 1-hour AI block in everyone's calendar, and ask managers to protect it.
- Week 3 to 4TightenHold the first monthly live session on a task your own team does every week, and start weekly office hours with the champion.
Month 2
First loop
Ship 2 workflows, time them, then ship 2 more, so the team sees the loop work.
- Week 5StartCollect use cases from every rep, give each an hours-saved estimate, and rank the list.
- Week 5StartTime the top 2 workflows the way they run today, so there is a before number.
- Week 6 to 7StartShip the first 2 workflows with the people who do the work, and time them again after a week of use.
- Week 7TightenPost both wins company-wide: what changed, the steps, and the hours saved.
- Week 8StartSplit training into 3 tracks: starters, daily users and builders. Ship the next 2 workflows from the ranked list.
Month 3
Hackathon
Run a 2-week hackathon so ideas come from the whole team, not only the champion.
- Week 9StartAnnounce a 2-week hackathon. Teams of 2 or 3 pick a task from the ranked list or bring their own.
- Week 10 to 11StartTeams build during their blocked AI time; the champion runs extra office hours.
- Week 11StartDemo day. Judge each entry: 50% hours saved, 30% how many people would use it, 20% how easy it is to reuse.
- Week 12StartShow leadership the 90-day total: hours saved per workflow, before and after. Make it a monthly report.
- Week 13StartScore this card again. Put the 3 lowest lines at the top of the next 90 days.
Start: the line scores 0 or 1. Tighten: it scores 2. Keep: it scores 3, so the step only checks it still holds.
The 10 lines
| # | Line, and what your score means | Score |
|---|---|---|
| 1 | One named person owns AI adoption, with hours for it in their weekA named owner has protected hours every week, and everyone knows who it is. | 3 / 3 |
| 2 | One person from each department meets every 2 weeksThere are department reps, but they meet when someone remembers. | 2 / 3 |
| 3 | A 1-page AI policy is pinned where people workOne page says what data may go where and which tools are approved, and it is pinned in the main channel. | 3 / 3 |
| 4 | Live training every month, on your own workflowsTraining happens a few times a year, or uses generic examples. | 2 / 3 |
| 5 | Training comes in 3 tracks: starters, daily users and buildersThere is one level, pitched at beginners. | 1 / 3 |
| 6 | Sessions are recorded into a library that is checked every monthThere is a shared library, but old or broken material stays in it. | 2 / 3 |
| 7 | A list of AI use cases, ranked by hours savedThere is a list, but nobody ranks it. | 1 / 3 |
| 8 | Wins are posted where the whole company sees themWins are posted, but without the time saved or how it was done. | 2 / 3 |
| 9 | Time for AI work is blocked in calendars every weekSome people block time for it on their own. | 1 / 3 |
| 10 | Hours saved are measured and shown to leadershipNobody knows what AI saves. | 0 / 3 |
The AI champion
Month 1 starts by naming this person. Print the role card.
Role card
AI champion
Who fits
- Curious: tries new tools without being asked, and says what did not work.
- Close to the work: knows how the team spends its day, not only what the slides say.
- Trusted: people already ask them how to get things done.
- Plain-spoken: can explain a tool to a new hire without jargon.
Not the most technical person in the company. The champion makes AI normal for everyone; engineers can help them build.
The 5 jobs
| Try one new tool or feature, and give it a verdict: adopt, watch or skip. | Every week |
|---|---|
| Run the live training session on a real task from the team. | Every month |
| Keep the session library current: add new recordings, remove stale ones. | Every month |
| Rank the use-case list by hours saved with the department reps. | Every 2 weeks |
| Hold office hours where anyone can bring a task and leave with a way to do it. | Every week |
About 5 hours a week, written into the role and protected by their manager.
A self-score: you pick 0 to 3 on each of the 10 lines, and the total runs 0 to 30. Bands: 0 to 9 You bought seats, 10 to 19 A start, no engine, 20 to 26 Almost there, 27 to 30 AI native. The 90-day plan is a fixed spine; each step is marked start (line at 0 or 1), tighten (2) or keep (3). First moves are the 3 lowest lines, earliest line first when tied. Nothing is fetched or stored: the share link holds only the 10 numbers and the team name you typed.