The Agentic Maturity Model
Vishal Sachar
Co-Founder & CEO of CLRT
The most common question a business owner asks about AI is also the hardest one to answer honestly: are we behind? It is unanswerable without a map, because "behind" implies a path, and most people have no picture of the path. So here is the path. Every organisation is somewhere on a five-step climb from AI curiosity to AI autonomy, and knowing your step is worth more than any prediction about where the technology is headed.
The five levels run from zero to four. At level zero there is curiosity and little real use. At level one, individuals reach for chatbots ad hoc, helpfully but without any system. At level two, AI sits inside defined workflows under close human supervision. At level three, agents run whole tasks on their own, with humans holding the gates that matter. At level four, systems run continuously and verifiably, with people moved up to strategy and exceptions rather than execution.
The instinct on seeing a ladder is to want the top of it. Resist that. For most companies, the largest gains are in the early climb, the move from scattered individual use to a first properly structured, supervised workflow. The model is not a race to level four. It is a tool for finding the one rung above where you stand, and stepping onto it.
That is what makes a maturity model useful rather than merely impressive. A framework you can locate yourself on is one you can act on.
A deeper dive
What actually changes between levels is not the model, it is the system maturity around it, and each step has a characteristic discipline that gates the next. The climb from level one to level two is about written process: you cannot put AI inside a workflow you have never defined, so the work is articulating how things are actually done. The climb from two to three is about verification and trust, because you only delegate what you can check, which means the prerequisite for autonomy is a reliable way to catch errors, a separate checker the worker does not control (the principle behind The Maker and the Checker). The climb from three to four is about governance and observability, because you only automate at scale what you can see, audit, and stand behind. This is why the levels are sticky and cannot be skipped. Each rung's discipline is the foundation the next rung stands on, and a company that tries to leap to autonomy without the verification underneath it does not reach level four, it reaches an unmonitored level one wearing a costume.
The sequence
- L0
Curiosity
Interest in AI but little real use.
- L1
Ad hoc use
Individuals reach for chatbots helpfully, without any system.
- L2
Supervised workflows
AI sits inside defined workflows under close human supervision.
- L3
Delegated tasks
Agents run whole tasks alone, with humans holding the gates that matter.
- L4
Verifiable autonomy
Systems run continuously and verifiably, people moved up to strategy and exceptions.
Work with CLRT
Where is your business actually standing on the climb? The CLRT Ascent diagnostic places you precisely and names the next rung. Find your level rather than guessing at it.
Vishal Sachar
Vishal Sachar is the Co-Founder and CEO of CLRT, where he helps UAE businesses make sense of applied agentic AI and put it to work. He writes on agentic systems, AI governance, and the economics of automation. Reach him at vishal@clrtstudio.com or on LinkedIn.
