A seven-level model assesses AI proficiency in the workplace

In 2026, American consultant Harrison Painter from LaunchReady.ai introduced a model that divides professional AI use into seven levels. The spectrum ranges from a first chatbot experience to the ability to coordinate entire systems, serving as a practical guide for professionals to understand their current stage. Although it is not an academic standard and psychometric validation is still ongoing, the proposal already serves as a reference for self-assessment.

From basics to orchestration: the seven stages

Levels 1 to 3: interaction with the tool

At level 1, the user has already experienced AI but formulates requests in a way similar to an internet search, receiving inconsistent responses.

At level 2, the user begins to specify context, format, and constraints. Instead of saying “summarize this report,” they indicate the target audience, the size, and the type of summary desired. The responses improve, but the tool is still seen as a mere request mechanism.

At level 3, the professional questions the answers, asks for alternatives, and insists when the result is unsatisfactory, instead of abandoning the tool. The first response is now considered a draft to be refined.

Levels 4 and 5: system management

Level 4 is marked by context engineering. Very long conversations tend to reduce the quality of responses, and those who master this stage know when to start a new session and what information to bring to it.

At level 5, the user stops using AI just for themselves and starts creating solutions for colleagues, defining what data the tool can access. An example is a human resources analyst who develops an assistant to clarify doubts about benefits.

Levels 6 and 7: organizational integration

At level 6, the best processes are transformed into documented and reusable flows, with clear steps, inputs, and success criteria. This requires the ability to convince other areas and overcome internal resistance.

Level 7 connects multiple flows into pipelines that operate with minimal human intervention. At the highest levels, technical skills give way to human competencies, such as leadership and the ability to drive change.

Where most professionals stand

Painter notes that most professionals place themselves between levels 2 and 4 when doing a first self-assessment, and that many overestimate their proficiency at one level. The observation is from the model’s author, not from an independent survey.

Additional data comes from Anthropic’s AI Fluency Index, which analyzed 9,830 conversations on Claude.ai in January 2026. In interactions where the user refined the result instead of accepting the first answer, an average of 2.67 additional fluency behaviors were recorded—almost double the 1.33 observed in other conversations.

The same study showed that the more polished the material generated by AI, the less likely the user is to check for errors.

How to determine your level

LaunchReady offers a free self-assessment in English that lasts less than ten minutes and is interrupted when the participant reaches their limit, according to the company. The result indicates the current stage and the next step in learning.

For those at the initial levels, the model suggests better structuring requests and reviewing responses. Professionals at levels 3 and 4 are advised to document flows that can be reused by others.