Skip to content

AI Maturity: Which Stage is Your Company In and What to Do Next

Most companies are experimenting with AI; few use it systematically to transform operations. The gap is not technology but maturity. Understand the five stages and what it takes to advance.

Marlos Carmo

Founder of Tolky

Updated on November 15, 2026

8 min read
AI Maturity: Which Stage is Your Company In and What to Do Next

Share

The impact of generative AI on results is still uneven. In McKinsey's "The state of AI" research (survey run in July 2024 with 1,491 participants), more than 80% of respondents did not see a tangible impact of generative AI on the company's EBIT, and only the so-called leaders attributed more than 10% of EBIT to generative AI (McKinsey, The state of AI). In another McKinsey survey (January 2025), only 1% of executives described their generative AI roll-outs as "mature" (McKinsey).

What separates those already reaping results from the rest is not access to better technology: the available tools are essentially the same. What separates them is operational maturity: the ability to turn AI experiments into reliable systems that run in production, deliver measurable results and improve over time.

AI maturity cannot be bought with a platform. It is a progression through distinct stages, each with its own challenges, blockers and natural next steps. If you prefer to start with a self-assessment, take the AI maturity test.

The five stages of the AI maturity model

StageMain characteristicsMinimum observable evidenceTechnology focus
1. CuriousIndividual, uncoordinated use of tools (for example, ChatGPT)No project with an owner and a metricWeb tools for personal use
2. ExperimentingOne or two pilots, run by IT or one areaPilot with documented results, still without financial targetsChatbots, content generation, knowledge bases (RAG)
3. PilotingOne project in real production, dependent on a championReal volume, real users and financial metrics trackedAgent integrated with systems and channels
4. ScalingMultiple agents and established governanceResults reported to the board regularly, a business owner per initiativeAgent orchestration, ERP and CRM integrations
5. OptimizingAI integrated into the core of the operation and continuous improvementNew use cases identified from production dataContinuous learning, global governance

Check your stage in the AI maturity test.

Stage 1: Curious

How to identify it: the company tests AI tools individually and without coordination. Employees use ChatGPT personally. Some manager ran a chatbot pilot that never scaled. There is enthusiasm, but no strategy.

What is happening: the organization is calibrating what AI can and cannot do. That is healthy. The risk is not experimenting too much, it is staying too long at this stage.

Typical blockers: no clear use case, internal skepticism or no executive sponsorship.

What to do next: identify a specific business process with high volume, available data and a clear success criterion (not the most complex one, the most immediate). Set up a project of about 60 days (a planning assumption) with a defined objective and result metric. The article on process automation with LLMs: real use cases in B2B companies lists cases with immediate impact.


Stage 2: Experimenting

How to identify it: one or two AI projects underway, usually a customer service chatbot pilot or AI in marketing. Projects are managed by IT or a specific area, with no regular executive visibility.

What is happening: the company has moved past curiosity and produces evidence that it works, but the projects are still technical experiments, not strategic initiatives.

Typical blockers: difficulty scaling beyond the pilot, resistance from other areas and poorly defined success metrics.

What to do next: turn the pilot into a business initiative: define the owner in the business (not just in IT), set financial metrics (cost per ticket, deflection, time saved) and present results in ROI language. The AI automation ROI framework helps build that calculation.


Stage 3: Piloting

How to identify it: the company has an AI project in real production, with volume, real users and performance data. Results are visible, but the initiative still depends on an internal champion.

What is happening: it is the most critical stage of the journey, and where many companies stop, not because AI does not work, but because the organization was not prepared to absorb the operational change it demands.

Typical blockers: resistance from the affected team (fear of replacement), unclear governance (who is responsible for the agent's quality?) and no continuous improvement process.

What to do next: institutionalize. Name an operations owner for the agent (not just IT), create a monthly performance review and expand the scope to a second process.


Stage 4: Scaling

How to identify it: multiple AI agents in production across different departments, with an established governance model and results reported regularly to the board.

What is happening: the company has left project mode and entered operating mode. AI stops being a digital transformation initiative and becomes part of how the company operates.

Typical blockers: fragmentation (each department built its own agent in isolation), no sharing of learnings and rising maintenance cost without standardization.

What to do next: build a central structure that governs standards, shares knowledge bases and coordinates initiatives (not necessarily an in-house AI team).


Stage 5: Optimizing

How to identify it: AI is integrated into the core of the operation. Agents learn from the feedback of every interaction, and the company uses the intelligence they generate to improve products, processes and strategy. New use cases are identified continuously.

What is happening: the company accumulates data, operational experience and governance maturity, which tends to make an advantage harder to copy. How long a competitor would take to replicate it varies by case.

Focus at this stage: continuous discovery of new use cases, optimization of agents based on production data and integration of new capabilities (multimodal, autonomous agents, predictive analytics) as they become viable.

Team mapping priorities with sticky notes on a wall: AI maturity requires a roadmap built together, not a tool purchase

How to know which stage your company is in

The fastest diagnosis combines three questions:

1. Do you have AI running in production with relevant volume and financial metrics tracked regularly?

  • No → Stage 1 or 2
  • Yes, but in a single project → Stage 3
  • Yes, in multiple projects with governance → Stage 4 or 5

2. Is there a clear business owner (not just IT) for each AI initiative?

  • No → probably Stage 1 or 2, regardless of what was implemented
  • Yes → a sign of Stage 3 or above

3. Does the board discuss AI results in financial terms at least quarterly?

  • No → the initiative is not institutionalized, even if technically advanced
  • Yes → a sign of Stage 4 or 5

For a fuller diagnosis, use the AI maturity test and measure your service efficiency in the agent efficiency test.

What keeps most companies from advancing

The transition from Stage 2 to Stage 3, from experimenting to having it in production, is the bottleneck where many companies get stuck. The blockers are predictable:

Choosing the wrong use case. Projects that try to solve the most complex problem first rarely reach production. Starting with the simplest high-volume case tends to reach production faster and justify the next project.

Lack of structured data. An AI agent is only as good as the knowledge available to it. Companies with no process documentation, up-to-date FAQ base or structured interaction history cannot train an effective agent and blame the AI when the problem is the lack of data.

No business owner. Implementations managed only by IT optimize technical metrics and ignore business metrics.

Expecting perfection before production. AI agents improve with use, and pre-production perfectionism is one of the main killers of AI projects.

The role of a platform in the maturity progression

Companies that build AI agents in-house, with their own engineers, language-model APIs and custom integrations, tend to take longer to reach the first agent in production (the timeline varies a lot by company) and later find that maintenance, updates and monitoring require a dedicated team.

A platform like Tolky compresses part of that cycle. Instead of building the infrastructure, the company focuses on configuring the agent for its use case, with connectors for common systems, monitoring tools and support for channels such as WhatsApp, webchat, Microsoft Teams and voice (other channels are coming soon). Setup is quick, and plans are published at /precos, from free Individual to Enterprise on request; for implementation timelines in your context, ask the team. The criteria for choosing a platform are in the guide on enterprise AI automation platforms, and anyone who wants to compare options can see /alternativas.


AI maturity is not a state you declare. It is a state you demonstrate with agents in production, financial metrics tracked and business decisions that change based on the intelligence the agents generate.

Where is your company on this journey, and what is the next concrete step? Take the AI maturity test or talk to our team to identify the use case with the greatest impact potential for your organization's current stage.

Share

Tags

  • stages of corporate AI adoption
  • AI roadmap for large companies
  • digital maturity diagnosis with AI
  • how to advance in the B2B AI journey
  • artificial intelligence maturity in companies

About the author

Marlos Carmo

Founder of Tolky

Marlos Carmo is an AI entrepreneur and founder of Tolky, the conversational-era infrastructure and AI CRM that unifies intelligent service, multi-channel support (such as WhatsApp and voice), live CRM, and operational intelligence in a single ecosystem. He is a finalist for the SXSW Innovation Awards and a member of Francesco's Economy, a global network of young entrepreneurs focused on innovation and social impact. He works connecting Artificial Intelligence and digital transformation in projects for large organizations.

Next step

Want to bring this to your operation?

Tati, Tolky's AI assistant, understands your scenario in a few questions, suggests the best path and connects you with someone from the team, with full context.

Read also