📈 AI Maturity Assessment (scale 1–5)
Every assessment starts with a question: where the company stands today? The Factory provides the BCG maturity curve—five stages—and a score from 1 to 5 across seven dimensions. Without this snapshot, the roadmap and ROI are up in the air.
💡 The 7 dimensions evaluated
Strategy & Vision · Data & Infrastructure · Technology & Tools · Talent & Skills · Governance & Ethics · Culture & Change · Use Cases & Value. Each receives a score of 1-5 with research evidence — it’s what makes the diagnosis defensible, not a guess.
5 stages
rating 1–5
readiness
of the research
⚡ Quick Wins (effort × impact)
A diagnosis alone doesn’t sell. What unlocks the budget is the quick proof of value: AI initiatives that run in less than 60 days, cost little, and show ROI. The effort × impact matrix prioritizes what to tackle first.
✓ It’s a Quick Win when
- ✓Implement in < 60 days
- ✓Investment < US$50K (or internal resources)
- ✓Low risk and minimal prerequisites
- ✓Demonstrable and measurable ROI
✗ NOT a Quick Win when
- ✗Depends on a long data migration
- ✗Needs a custom-trained model
- ✗Requires team reorganization
- ✗Value is vague and has no clear KPI
short timeframe
low cost
little uncertainty
proof of value
⚖️ Build vs Buy
For each use case, a million-dollar question: build or buy? The framework doesn't decide based on preference — it weighs factors and gives a score. The decision can be defended before the board.
// weighted decision matrix (weight × score)
| Factor | Weight | Build | Buy |
|---|---|---|---|
| Time to value | 25% | slow | fast |
| Total cost (3 years) | 20% | variable | predictable |
| Customization requirements | 15% | total | limited |
| Internal capacity | 10% | requires a team | outsourced |
| Lock-in risk | 5% | low | high |
✓ Tends to BUY when
- ✓Needs quick value
- ✓The use case is common (not a differentiator)
- ✓Doesn’t have an in-house ML team
→ Tends to BUILD when
- →It’s a strategic competitive advantage
- →Customization is extensive and the data is sensitive
- →Want to avoid vendor lock-in
weight × score
actual cost
not wholesale
output risk
💵 AI ROI
The deliverable that convinces finance. Numbers speak louder than pretty slides. The framework starts with four sources of value and calculates simple ROI, payback, and 3-year NPV—with the assumptions always explicit.
⏱️ Efficiency gains
Hours and costs saved, fewer errors. The easiest gain to measure.
📈 Revenue Increase
Conversion, business speed, customer retention.
🛡️ Risk reduction
Avoided compliance, security, and operating costs.
🚀 Strategic value
Competitive advantage and capacity to innovate.
// the three figures every client wants to see
ROI simples = (Benefícios − Investimento) / Investimento × 100 Payback = Investimento total / Benefício mensal → em meses VPL (3 anos) = Σ fluxo / (1 + 0,10)^ano (taxa de desconto 10%)
💡 Practical tip
Always deliver three scenarios: conservative, base, and optimistic (sensitivity analysis). Shows you’ve thought about risk—and protects you from the question "what if it doesn’t work?".
of value
in months
3 years, 10%
sensitivity
🗺️ 30/60/90/180/360 roadmap
It’s what turns "use AI" into an actionable plan. Five time-based phases, each with a clear objective, activities with an owner, and verifiable milestones. The client opens it on Monday and knows exactly what to do.
Foundation
AI committee, data audit, prioritizing quick wins, vendor shortlist.
Quick Wins
Launch 2-3 initiatives, get the first tool live, and capture the first ROI metrics.
Scale
Expand successful pilots, 3+ initiatives in production, Center of Excellence taking shape.
Optimize
Refine implementations, expand to new departments, launch the executive dashboard.
Transform
AI at the core of processes, measurable competitive advantage, self-sustaining capability.
💡 Leading vs. lagging indicators
Track both: leaders (completion rate, adoption, training) predict the future; overdue (ROI, efficiency gains, revenue impact) prove the past. Only leaders turn it into theater; only laggards make it a surprise.
30 → 360 days
with checkbox
by activity
leader + behind schedule
🛡️ Data governance & policy
It’s the deliverable that puts legal and the board at ease. Without governance, no pilot makes it to production. It defines what can turn into AI training, input, and output — and who is responsible for it.
// data classification and AI usage
| Level | AI training | AI Input | AI Output |
|---|---|---|---|
| Audience | allowed | allowed | allowed |
| Operational | allowed | allowed | allowed |
| Client / PII | prohibited | restricted | review |
| Transaction | with consent | anonymized | allowed |
✓ A good policy has
- ✓Clear classification (Public → Restricted)
- ✓Defined roles (CDO, data stewards)
- ✓Compliance mapped (LGPD, GDPR, CCPA)
- ✓Acceptable use policies for tools
✗ Signs of weak governance
- ✗PII dumped into a public AI tool
- ✗No one owns the data
- ✗No audit trail for decisions
- ✗"We’ll deal with it later" — and they never do
4 levels
clear rules
data owner
compliance
✅ Module summary
🎯 Mission 2.1 — Quick Wins for an industry
Choose an industry you know (retail, clinic, law firm…) and apply the framework:
- List 3 quick wins AI for them (<60 days, low cost).
- For each one: department, effort (low/medium), impact, and a KPI.
- Rank the three in a mini effort × impact matrix.
Success: 3 quick wins prioritized. What you gained: the outline of the quick wins deliverable—ready to paste into a real assessment.
Next module:
2.2 — Lazy RAG: huge PDF → cheat sheet (how to distill the playbooks that power these frameworks)