AI Training Week 6: Champions and Final Projects

After watching dozens of 6-week AI programs end, my conclusion is this: week 6 is not a graduation ceremony — it's the handover that determines whether the program compounds for 6 months or evaporates in 6 weeks.

Why week 6 is mostly about handover, not finale

The trap most companies fall into is treating week 6 as a celebration of week 5's work. That produces a polished demo and a quiet death by week 10. Microsoft's 300,000-employee Copilot rollout had a great launch and lost over 80% of usage in three weeks, because the operating handover after launch was thin. Atos, by contrast, scaled from 300 early Copilot licenses to 15,000 trained employees by treating champions as a permanent role with allocated time — that's the bridge pattern that holds.

About 89% of users who push past the productivity dip stay active 20 weeks later, per Microsoft's internal data. Week 6's job is to make sure your team gets past that dip — not to give them a slideshow.

Definition: Operating plan — a 90-day document that names what each role-track will ship, what each champion will run, and what gets reviewed at week 9, 12, and 18. Different from a roadmap.

What week 6 should actually contain

The structure that keeps adoption alive:

1. Monday — 60-minute final-project kickoff. Each role-track names the one project they'll ship by Friday. Built on agents from week 4 + risk frame from week 5.
2. Tuesday/Wednesday — async build + polish. Pairs finalize their projects. Champions on call.
3. Thursday — 90-minute champion graduation. Formal handover: title, time allocation (typically 10-20% of work week), reporting line, monthly stipend or recognition.
4. Friday — 90-minute company finale + 90-day plan. Each role-track demos final project AND presents 90-day operating plan. Founder approves.

Two artifacts ship on Friday: the project demo and the operating plan. If only the demo ships, you've done a graduation. If both ship, you've done a handover.

Definition: AI champion (post-graduation) — formally an employee allocated 10-20% of their work week to running their role-track's AI program: office hours, lab refreshers, new-hire onboarding, agent maintenance.

The final-project brief (copy/paste)

Each role-track ships ONE final project on Friday.

The project must:
1. Build on the role-track's week-4 agent OR ship a new agent of similar scope.
2. Pass week-5 Responsible-AI playbook checks.
3. Have a named owner (usually the champion) for the next 90 days.
4. Have a metric: time-saved per week, deflection %, throughput, or output volume.
5. Have a 90-day plan: what gets measured at week 9, 12, 18.

Friday demo: 5 minutes for the project + 5 minutes for the 90-day plan.
Founder approves both, or sends back to revise.

The 5-minute 90-day plan is the part most cohorts try to skip. Hold the line.

Tool tip (Course for Business): The reason Augment, don't replace has to be reaffirmed in week 6 is that this is when leadership starts asking "so where are the headcount savings?" The honest answer for most SMBs is: there aren't direct ones in 6 weeks; there are 70 person-hours/month worth of saved time per agent that role-tracks can reinvest into work that wasn't getting done. The 6-week program at https://course.aiadvisoryboard.me/business closes week 6 with a board-ready 90-day plan that frames AI as capacity expansion, not headcount reduction. (Course for Business)

The 90-day operating plan (template)

Role-track: [name]
Champion: [name]
Champion time allocation: [%]
Reporting line: [who reviews monthly]

WEEK 9 REVIEW
- Final project still running? Y/N (with metric snapshot)
- New agents shipped: [count]
- Office hours attendance: [count]
- Top blocker: [one sentence]

WEEK 12 REVIEW
- Cumulative time-saved (this role-track): [hours/week, ranges OK]
- New use cases identified: [count]
- Champion handover candidate identified? [Y/N]

WEEK 18 REVIEW
- Adoption rate (employees actively using AI weekly): [%]
- Cohort 2 readiness: which employees ready to be champions?
- Decisions for leadership: [list]

Three review points, named owners, named metrics. Anything more elaborate doesn't survive week 9.

Good vs bad week-6 outcomes

Bad: "We had a great demo day. The champions were applauded." Good: "Each champion has 15% allocated time, a reporting line, and a 90-day plan. Week 9 review is on the calendar."

Bad: "We hit our target of 5 shipped agents." Good: "We hit 5 agents, plus week-9 review will check whether they're still running and what we'll add by week 12."

Bad: "Champions volunteered without time allocation." Good: "Champion time is in the operating budget. Their managers know. Performance reviews include AI-program contribution."

The good versions treat week 6 as the start of the next 90 days, not the end of 6 weeks.

Team scan (what AI champions report after week 6)

• Cohorts that allocate champion time formally retain over 80% of week-6 use rate at week 12; cohorts that don't, drop below 40%.
• The single most common champion failure mode is "I'm doing this on top of my real job" — and the manager hasn't been told.
• About 1 in 3 role-tracks identifies a successor champion candidate by Friday — that's the pipeline for cohort 2.
• Final-project metrics that work: time-saved (hours/week), deflection %, throughput. Metrics that don't: vague "productivity" claims.
• Founders consistently underestimate how much the formal champion handover signals to the rest of the company.
• Champions report that the 90-day plan is what makes them feel safe to push back when their manager wants to pull them off the program.
• Cohorts whose week-6 demos go to the board produce visibly better follow-through than cohorts whose demos stay internal.
• Around 1 in 4 final projects gets meaningfully expanded scope at week 12 — usually because data access improves once people see the output.
• The cohort Slack should NOT be archived after week 6 — it's the connective tissue between cohort 1 and cohort 2.
• The AI Tax (~37% of saved time re-spent on rework if training was poor) shows up most visibly at week 9 — that's the early-warning signal.

Micro-case (what changes after 7-14 days)

A 220-person logistics SMB I advised closed week 6 with five role-tracks, five final projects, five graduated champions — and most importantly, a board-presented 90-day operating plan with named time allocation. By day 14, four of the five champions had each held one office-hours session for non-cohort employees ("cohort 2 prep"). The CEO included AI-program contribution as a line in the next quarterly review cycle. By week 12, the operations role-track had shipped two more agents on top of the week-6 final, and time-saved across the team was tracking around 70 person-hours/month — consistent with patterns from peer companies. Compare to a peer firm that ran the same 6-week program but skipped the formal champion graduation: by week 12, three of five champions had quietly stopped running office hours, two of five projects had stopped running, and the head of L&D was starting from scratch with a "phase 2" pitch.

Note on this case: This example is illustrative — based on typical patterns we observe with companies of 30-500 employees, not a single named client. Specific numbers are rounded approximations of common ranges, not guarantees.

Tool tip (Course for Business): The hardest part of week 6 isn't building the final project — it's making the case for permanent AI Champions (1:15-20) allocation when leadership wants to "see how it goes" before committing time. The case to make: champions cost ~10-20% of one employee per 15-20 staff, and the alternative is the Microsoft pattern of usage collapse within weeks. The 6-week program at https://course.aiadvisoryboard.me/business hands off a board-ready operating plan that includes the time-allocation case explicitly, because 80% of post-program adoption depends on it. (Course for Business)

FAQ

How much time should champions get formally? For most SMBs, 10-20% of a normal work week — roughly 4-8 hours. Less than 4 and they can't run office hours plus maintenance. More than 20% and you've created a different role.

What if a champion doesn't want the formal role after week 6? That happens in about 1 in 5 cohorts. Don't force it — name the next-best candidate and re-graduate at week 9. Forced champions kill cohorts.

Should champions get extra pay or just recognition? Recognition + time-allocation works in most SMBs. A small monthly stipend or one-off bonus signals seriousness. Avoid making it a percentage of any cost-savings claim — that distorts incentives.

Does week 6 need to involve the board? For 30-200 person companies, founder/CEO sign-off is usually enough. Above 200, including the board (or whoever holds budget) at the week-6 demo materially improves follow-through. (We separately have a daily-management advisory product on aiadvisoryboard.me, but week 6 should stay in-house.)

What's the metric that proves week 6 worked? At week 12, are champions still running scheduled office hours? Are final projects still running? If both yes — week 6 worked. If either no — diagnose at week 9, don't wait.

Conclusion

Week 6 isn't the end of an AI program — it's the start of the operating phase. Final projects ship, champions graduate into formally-allocated roles, and a 90-day operating plan goes to the founder or board. Skip the handover and the program evaporates; do the handover and the next 6 months compound. After week 6, your company doesn't have an "AI training program" anymore — it has a way of working.

Next step: lock champion time allocation in writing before Friday's demo, and put the week-9 and week-12 reviews on the calendar today.

If you want every employee to ship their first AI automation in five days — book a 30-min call and we'll map your team's first week: https://course.aiadvisoryboard.me/business