Maturity models exist for a specific purpose: create shared language about where the organization is today and where it needs to go. Using them as a trophy ("we need to be level 5") is the mistake. Using them as a mirror ("we're at level 2 — what does that mean for the project?") is the work.
The 5 maturity stages—each step requires the foundation of the previous one.
🔍 What maturity models are
Maturity models describe progressive stages of organizational capacity. In AI, the most widely used approach in Brazil is inspired by data maturity frameworks and CMMI — it ranges from ad hoc operations (level 1) to AI as a core competitive advantage (level 5).
🧭 A mirror, not a trophy
The model is a diagnostic tool, not a competitive ranking. An organization at level 2 isn’t "worse" — it’s exactly where it should be given its history and investments. The goal is to reach level 3, not pretend to be at level 5.
- •Error: leadership wants to "get to 5 in 6 months" without a basis in the earlier stages.
- •Correct: identify the current level honestly and propose the next feasible step.
shared language
where we are today
realistic next step
sequenced roadmap
1️⃣ Stages 1 and 2 — Ad hoc and Aware
Most Brazilian SMBs are in this range. This is where consultants find the most foundation preparation — and where most premature AI projects fail by skipping this foundation.
Level 1 — Ad hoc
Most SMBs
Isolated AI experiments by motivated individuals. No structured data for AI, no clear vision, no dedicated investment. What works, works by accident — nothing is replicable or scalable.
Consultant recommendation: inventory data, document critical processes, and identify the simplest first use case.
Level 2 — Aware
Engaged leadership, no structure
Leadership understands that AI is important. There are discussions, perhaps a pilot project, but no dedicated budget, no team, and no structured data to support a real project.
Consultant recommendation: help build the business case for the first project with clear ROI—it turns “we think it’s important” into a budget commitment.
3️⃣ Stage 3 — Active
Stage 3 is the turning point: the organization has projects underway, initial results documented, and a foundation for expansion. This is where a consultant can have the greatest impact — and where the risk of getting the strategy wrong is highest.
⚡ Signs that the organization is at level 3
- ✓At least 1-2 AI use cases in production with defined KPIs
- ✓Data structured for the use case being executed
- ✓Technical team (internal or outsourced) with clear responsibility
- ✓Dedicated budget approved for AI
💡 What the consultant does at level 3
Helps expand what works (replicating the winning use case in other processes), establishes basic governance before scaling, and documents lessons from the first projects as a foundation for the long-term roadmap.
4️⃣ Stages 4 and 5 — Operational and Transformational
The more advanced stages require robust internal capability. Organizations at this stage no longer rely on external consultants to operate—they rely on them to innovate and handle specialized cases.
Level 4 — Operational
- • AI in multiple critical processes
- • MLOps: automated deployment, monitoring, and retraining
- • AI KPIs tied to business KPIs
- • Formal governance with documented policies
- • Established internal data/AI team
Level 5 — Transformational
- • AI as a core competitive advantage
- • New business models enabled by AI
- • Data valued as a strategic asset
- • AI in the company’s core product/service
- • Internal research and innovation capability
📌 Brazilian market realities
The overwhelming majority of Brazilian companies are at levels 1–3. Few mid-sized businesses reach level 4. Level 5 is reserved for digital-native technology companies and large corporations with years of investment. Calibrating expectations to this reality is part of a consultant’s duty to be honest.
⚠️ Pitfalls of the maturity model
The maturity model is useful when used correctly. When misused, it can create more problems than it solves — whether due to leadership's excessive ambition or the consultant's underestimation.
✗ Misuse of the model
- ✗Use as a benchmark (“our competitor is at 4; we want to get to 4”)
- ✗Jumping from level 1 straight to level 4 projects
- ✗Declare a higher level to impress investors
- ✗Treating the model as a checkbox instead of a real diagnosis
✓ Good use of the model
- ✓Honest, evidence-based diagnosis, not perception
- ✓Goal = advance one stage, not jump to the top
- ✓Use as a basis for a sequenced roadmap
- ✓Revisit the diagnosis every 6 months
🗺️ Use the model as input for the roadmap
The maturity model is the input, not the output. The output is the roadmap sequenced in waves — each wave takes the organization one level higher, building on the previous level’s foundations.
🔗 From evaluation to the next step
💡 Practical tip
Present the current level as the result of a shared assessment — not the consultant’s verdict. When the client reaches the conclusion with you, they’re much more likely to accept the recommendation.
🎒 Module summary
Next module:
3.3 — Data and infrastructure audit: inventory, quality, LGPD, and readiness for RAG/training