of companies attribute some EBIT impact to their use of AI — practically the same level as the year before, even with more companies scaling the technology.
Source: McKinsey, The State of AI in 2026
Quick check
Check the ones that apply. It takes less than a minute and collects no data.
Want a complete view of how your company is managed? Take the free assessment: 15 questions, about 5 minutes →
Signs based on research by McKinsey (The State of AI) and Gartner.
Why so many AI projects stall
of companies attribute some EBIT impact to their use of AI — practically the same level as the year before, even with more companies scaling the technology.
Source: McKinsey, The State of AI in 2026
main reasons for abandoning generative AI projects after proof of concept: poor data quality, inadequate risk controls, rising costs and unclear business value.
Source: Gartner, July 2024
The companies that gain the most from AI redesign their workflows around it, instead of just fitting it into the process that already exists.
Source: McKinsey, The State of AI in 2026
That is exactly where Mapa comes in: before the tool, we organize the process, the data and the usage rules. Then we implement and measure.
What we do
Clear rules for using AI safely: who can use it, for what, with which data and how to measure it.
From use case to agent in production: we choose where AI pays off and implement it with indicators.
A structure to run digital initiatives as a portfolio — with priorities, owners and deadlines.
Reliable, accessible data — the foundation without which no AI works well.
In practice, not just in theory
Since 2023, Marcelo Neves, Mapa's founder, has led AI adoption at FECOMERCIÁRIOS-MG: AI agents and automations designed, built and orchestrated with different models and platforms, now in use across several areas.
M.A.P.A. Method applied to AI
No. Organizing the data is part of the work. In the assessment, we identify what can already be used and what needs to be prepared first.
We are vendor-independent. We use models such as Claude, GPT and Gemini and platforms such as n8n, CrewAI and Copilot Studio — the choice depends on the use case, the cost and your company's data rules.
They are part of the project from the start: we define which data can be used, in which tools and with which controls, and we record this in the AI usage policy.
It depends on the use case chosen. We prioritize cases with fast, measurable returns so the first gain shows up early and funds the next steps.
Tell us about your challenge and talk directly with Marcelo Neves.