1. Explore
Each person tests on their own.
- What you hear
- "People use AI, but everyone in their own way."
- Next move: Decide
- Choose a real problem, understand the process and define who looks after the data.
- How we help
- Assessment in 2 weeks
AI Maturity Guide
AI maturity means repeating results. The DB1 Itênex view.
How we think about AI in companies, how we see each maturity level and why we build agents the way we do.
What we believe
Starting to use AI has become easy. The hard part is repeating the result when the problem, the team or the person who started it changes. To us, that is what AI maturity is: repeating results, with control, without depending on whoever started.
The talented cook
Makes an incredible dish. When the cook goes on vacation, the dish goes too. This is the company where one or two people work miracles with AI, and nobody else knows how.
The restaurant
Has recipes, suppliers, standards and a team. The dish comes out the same every night, and the next one on the menu comes easier. This is the company that turned AI into a way of working.
The question that sums it all up
Does the second case go better than the first, even without the person who drove the first?
The problem is rarely the technology.
AI amplifies what the company already is. A clear process gains speed. A confusing process makes mistakes faster. In our experience, when AI stalls, one of these four things is missing.
A decision
Nobody chose which problem to solve, or who answers for it.
An understood process
A flow nobody mapped got automated, with data nobody looks after.
A measure
Nothing was agreed up front. Afterwards, nobody can say whether it worked.
A way to keep going
Only one person or one vendor understands how it works. When they leave, everything stops.
Others who see it this way
"AI doesn't fix a team; it amplifies what's already there."
DORA, 2025 reportLeading companies start with clarity on where AI creates value, redesign the work and measure the effect on results.
BCG, September 2026Over 40% of agentic AI projects are expected to be canceled by the end of 2027, due to cost, unclear value or inadequate risk controls.
Gartner, June 2025 forecastOf the 103 IT professionals we surveyed at an event, 76% would only use AI-generated code with review before production.
DB1 Itênex, APIX 2026 survey
Four levels. Each one points to the next move.
The level is what the company already does repeatably. Having done it once does not count yet.
2. Decide
The company chose where to act and knows who looks after what.
- What you hear
- "We know the problem and who looks after the data. We still need to see if it works."
- Next move: Prove
- Run a controlled test, measure the value and agree when to continue or stop.
- How we help
- Pilot in about 30 days
3. Prove
The value was measured in a test, with criteria to continue or stop.
- What you hear
- "The pilot delivered, and we have the numbers. Now it needs to become routine."
- Next move: Sustain
- Put it in production with security and support, and pass the knowledge to the team.
- How we help
- Measurement and governance to scale
4. Sustain
It works day to day, and the second case went better than the first.
- What you hear
- "We already run in production, with clear rules, and the team drives the next case."
- Next move: Repeat in more areas
- Take what works to other areas of the company.
- How we help
- Agent squad with your team
The sentences are examples to help recognize each level, not client quotes.
Take the maturity previewThe level shows where you are. The pillar shows what is holding you back.
That is why the preview points to your weakest pillar. It is usually the best place to start.
| Pillar | The question | What we do |
|---|---|---|
| Market and customers | Do we know where AI would create the most value? | We rank use cases by return and show where an agent pays off. |
| Resources and investment | Is there budget and people to build and maintain it? | We size what the first agent needs, and your team builds alongside ours. |
| Technology and infrastructure | Can AI query the data and systems? | We map what the agent will query and what needs fixing first. |
| Culture and leadership | Who drives the topic, and toward what goal? | We work with leadership to set an owner, a goal and a visible first case. |
| Processes and governance | Who controls what AI decides and does? | We write the scope and limit rules, with human approval and an audit trail. |
One view is not enough
Executives, managers and operations often see the same company differently. That is why a full assessment listens to all three.
Five principles guide how we build agents.
Each one exists to prevent one of the reasons AI stalls.
We start with the problem, not the tool.
First the process that takes the most time or the error that costs the most. Only then do we talk about technology.
We understand the process before automating it.
We map the flow, the data and the systems. What needs fixing comes before the agent.
We agree on the measure before building.
Accuracy, response time and cases handed to a person are defined before going to production.
People approve what matters.
Every agent follows written rules for scope and limits, with human approval and an audit trail.
The knowledge stays with your team.
Your team builds with us and keeps the architecture, the rules and the playbook.
And we tell you when it is not worth it.
If the process changes every week or depends on negotiation, an agent is not the answer yet. If an off-the-shelf tool does the job, we say that too.
Why do this with DB1 Itênex.
Because an AI agent in production is software engineering: it has to talk to your systems, follow rules and keep working after delivery. That has been our job for 26 years.
- 26 years of critical software engineering
- 4x faster than the traditional method, at 60% of the cost, in a portal migration at BRK
- 16M people depend on BRK's digital platform, which we work on
The people who sell are the people who build
Applied AI engineers write the agent with your team. No hand-off to a junior team after the sale.
We use what we recommend
Clovis Agentic and Harness are our own tools, used on our own projects, with the same written rules and human approvals.
Critical systems are our day-to-day
Connecting an agent to the ERP, the CRM or an old system with no API is routine for people who come from legacy and mission-critical work.
Certified governance
CMMI DEV/3, ISO/IEC 27001 and ISO/IEC 27701: access control, audit trail and data protection from day one.
Open method
You see how we work before you hire us. These materials are public:
No dependence on us
Architecture, rules and playbook stay in your environment. You take over whenever you want.
From people who work with us
"One thing that caught my attention from the beginning of the partnership with DB1 is the relationship, the way they engage with the client, always staying close, trying to understand what our needs are."
A case · Utilities and energy
From "where do we start" to agents in operation.
- ExploreThe company knew it needed AI but did not know where to apply it.
- DecideA maturity assessment, with use cases ranked by return.
- ProvePilot validated and results measured: 98% accuracy on the first deliveries.
- Toward SustainPurpose-built agents in production, inside the operation.
BRK, XP, Omni, Avenue, AL5 Bank and CNA are among our cases.
See the casesFrequently asked questions
The preview gave me a level. How much is it worth?
It is a starting point for the conversation. These are one person's answers, and the points and cutoffs are our choice. It is not a certification or a ranking. For a firm reading, an assessment listens to more people in the company.
Do I need to be at a high level to start?
No. Every level has a next move, and that is where we start. If the foundation is not ready yet, fixing it is the first job.
Do technology companies start ahead?
Not necessarily. Knowing technology is different from repeating results as an organization.
We have an AI pilot that stalled. Can it be salvaged?
Almost always. We assess the architecture, data and governance of what exists, identify why it did not scale and define the path to production.
Will we depend on DB1 Itênex afterwards?
No. Your team builds with us and takes over whenever it wants. If you prefer, a squad of ours keeps evolving the agents with you.