Start Here · 10-Minute Actions
What should you do next to unblock your AI transformation? Each row below distills one of the top 10 questions into what we see in Acme's baseline, the single highest-leverage move it points to, and why that move pays off. Tap any number to jump to the full answer and its evidence.
| # | What We See on the Scorecard | Why It Matters | |
|---|---|---|---|
| Adoption & People | |||
| 1 | Who's adopting? Not early career 69% responded with 50 concrete wins, but 31% said nothing and only one early-career person answered. Evidence: §1 · §2 · §3 ↓ | This Week Have managers personally reach out to the 23 non-responders and every early-career hire, listen first, don't lecture. |
The silent third is where stalled adoption and hidden risk live, and disengaged juniors are tomorrow's capability gap. |
| 2 | The gap is individual, not team 92% of maturity variation sits within teams, every team has power users and strugglers side by side. Evidence: §1 ↓ | This Month Launch peer learning inside every team, pair each team's strongest users with its strugglers. Cancel any plan aimed at "lagging teams." |
A team-targeted program would miss 92% of the real gap. Peer pairing hits the gap exactly where it lives. |
| 3 | Skills: yes. Capacity: no Deep experts are already building agents, after hours, on their own dime. Time is the binding constraint. Evidence: §2 · §3 · §4 ↓ | This Month Protect 2–4 hours per week per person for AI learning and sharing, and recognize it in performance goals. |
You can't buy your way past a time shortage. Unprotected learning burns out your best people, or walks out with them. |
| Value & Measurement | |||
| 4 | Real value, trapped in individuals A 2–3 person team cut a procurement cycle from 4–5 months to 6 weeks, but nearly every win is solo and unshared. Evidence: §1 · §2 ↓ | This Week Broadcast the procurement win as the house template, and require every reported win to be written up for reuse. |
Undocumented wins leave with the person who built them. One celebrated template turns retail value into wholesale value. |
| 5 | The three numbers that predict success Baseline AI IQ 58/100, with the two lowest dimensions, Process (2.84) and Visible Results (3.00), the ones that matter most. Evidence: §1 · §2 ↓ | This Month Put three metrics on the exec dashboard: Process Orientation, Visible Results, and the shared-win ratio (today: near zero). Schedule the re-survey. |
These are the numbers that separate organizational transformation from a pile of personal productivity hacks. |
| Blockers, Risk & Governance | |||
| 6 | The blocker is provisioning, not tech Staff are encouraged but not provisioned: paying for their own subscriptions, running out of tokens mid-project. Evidence: §1 · §2 · §4 ↓ | This Month Fund a central AI budget and enterprise-tier licenses. End out-of-pocket subscriptions this quarter. |
Encouragement without provisioning caps your ROI at whatever employees will personally pay for, 71% of teams cite budget/cost. |
| 7 | Shadow AI is the operating model The unsanctioned tool is out-cited 2-to-1 over the sanctioned one, your best wins run on personal accounts, near confidential data. Evidence: §2 · §4 ↓ | This Month Legitimize, don't crack down: issue enterprise accounts for the tools people actually use, plus plain-language data guidelines, within 30 days. |
Shadow AI concentrates in your highest performers. Enforcement would erase your best wins overnight; legitimization keeps them and closes the exposure. |
| 8 | Governance instincts are missing 82% of people are explorers; zero Optimizers or Observers, almost nobody wired to standardize or ask "should we?" Evidence: §3 · §4 ↓ | This Quarter Stand up an AI quality-and-risk review group from your 11 Stewards, Integrators, and Stabilizers, before expanding AI autonomy any further. |
As AI gets more autonomous, unchallenged outputs become unreviewed decisions. Build the check before the stakes rise. |
| Scale & Transformation | |||
| 9 | No path from pilot to scale, yet Nothing built on a personal account can be shared, and one colleague, cited by peers across 9 teams, is scaling AI for free. Evidence: §1 · §2 ↓ | This Month Formalize the champion role for your most-cited peer mentor, and stand up a shared repository of prompts, skills, and agents. |
Scaling is already happening informally through one person. Fund it, or lose the only bridge from pilots to organizational capability. |
| 10 | Work is already being redesigned Employees are compressing admin and drafting work and pulling outsourced work in-house, shifting from doing to orchestrating. Evidence: §2 · §3 · §4 ↓ | This Quarter Pick 2–3 roles where drafting work is compressing and formally redeploy the freed hours to strategy, partnerships, and judgment work. |
Redeployment, not reduction, is the play your own workforce is asking for, formalizing it turns quiet job-crafting into transformation. |
51 of 74 people responded and shared 50 concrete wins, from email drafting to multi-agent systems, indicating a broad grasp on AI possibilities. However, 23 people (31%) didn't respond, several who did left every field blank, and only one early-career person answered at all. What's up with the missing juniors?
92% of the variation in AI maturity sits within Acme's teams, not between them. Every team has power users and strugglers sitting side by side, so a program aimed at the lowest-scoring teams would miss almost all of the real gap. The right intervention is organization-wide peer learning, pairing each team's strongest users with its strugglers.
Acme's skills bench is unusually strong, deep experts are already building agents and internal tools without being asked. What's missing is capacity: time surfaces over and over as the thing people don't have. Staff learn after hours, fund their own training, and build on weekends. That's a provisioning gap which needs protected and incentivized time for learning and sharing, not more tools or training.
"I built my own agent to help me manage my emails. However, I did it after working hours, at work, there is no time I can spend on learning and building with AI." Staff member, Acme baseline
The value is real and sometimes dramatic, an AI-assisted procurement team of 2–3 people cut a country procurement cycle from 4–5 months to 6 weeks, and teams have pulled formerly-outsourced work in-house. But nearly every win is individual. Value is being created solo and rarely aggregated to the whoe company. Each person's AI capabilities walk out the door with them.
Acme now has a baseline, AI IQ 58/100, Level 3, and the two lowest dimension scores are the two that predict whether individual wins become organizational capability: Process Orientation (2.84) and Visible Results (3.00). We'd add one homegrown metric: the shared-win ratio, the fraction of reported wins used by more than one person. Today it's near zero. If those three numbers move at the next survey, transformation is working; if not, Acme is only accumulating personal productivity.
The technology works, Acme's people have proven it. The blocker is a management gap: staff are encouraged to use AI but not provisioned for it. They pay for subscriptions out of pocket, run out of tokens mid-project, and can't connect AI tools to the organization's own email and files. Buying more tools without fixing provisioning and process would change nothing.
"Very simply, I run out of tokens from my $20 plan that I pay myself, the organization could provide us with a higher tier." Analyst, Acme baseline
Shadow AI at Acme isn't a fringe risk, it's how the most impressive work gets done. The procurement acceleration, the agent frameworks, the dashboards: nearly all built on personally-funded accounts, outside official policy, sometimes adjacent to confidential data. And it concentrates in exactly the wrong place to crack down on: your highest performers. The fix is legitimization, enterprise licenses and clear data guidelines, not enforcement, which would erase your best wins overnight.
82% of Acme's people are Shapeshifters or Experimenters, roughly double the global benchmark, and there are zero Optimizers or Observers among established staff. Everyone wants to explore; almost no one is wired to standardize, quality-check, or ask "should we?" Acme's own people can feel it, an audit specialist worried colleagues "may not challenge AI responses enough." Before expanding AI autonomy, Acme needs to recruit, develop, or deliberately borrow governance instincts: its 4 Stewards, 4 Integrators, and 3 Stabilizers are the entire bench.
There is currently no path from pilot to scale at Acme, not because pilots fail, but because nothing built on a personal account can be shared, and a workforce of Experimenters keeps starting pilots rather than spreading them. The unlock is twofold: shared infrastructure (an enterprise AI environment plus a repository of prompts, skills, and agents, literally what employees asked for), and people, one design-team colleague was cited by peers more often than anyone else, across 9 teams. That person is Acme's scaling mechanism, currently working for free. Formalize the role.
Acme doesn't need to design this from a blank page, employees are already redesigning their own jobs. Admin and drafting work is compressing, formerly-outsourced creative work is moving in-house, and the most advanced users describe the same destination: moving from doing tasks to orchestrating systems that do them. The workforce play is redeployment, not reduction, the freed hours are being reinvested in strategy, partnerships, and judgment work.
"I want to shift my role from a developer of tools to an architect of automated pipelines." ML Engineer, Acme baseline
Scorecard: Evidence Map