One of the most common questions we hear from organisations is: what does good AI proficiency actually look like? Without clear benchmarks, it is impossible to set meaningful goals, measure progress, or compare your workforce against industry standards. At Inpromptify, we have developed a benchmarking framework based on assessment data from thousands of professionals across industries.
The Four Proficiency Levels
Our benchmarking framework defines four proficiency levels. Foundational users can perform basic AI tasks: simple prompting, accepting or rejecting AI outputs at a surface level, and using AI for straightforward tasks like drafting emails or summarising documents. They typically score in the 25th to 50th percentile on our assessments.
Competent users demonstrate solid working proficiency. They use advanced prompting techniques, can identify common AI errors including hallucinations, understand the strengths of different AI models, and integrate AI into their regular workflows. They score in the 50th to 75th percentile. Proficient users have deep practical skills across all five dimensions. They can architect complex multi-step AI workflows, critically evaluate outputs against domain expertise, select appropriate tools for different tasks, and train others. They score in the 75th to 90th percentile.
Expert users represent the top tier. They push the boundaries of what AI can do in their domain, build novel applications, understand the technical underpinnings well enough to diagnose and work around model limitations, and actively contribute to their organisation's AI strategy. They score above the 90th percentile.
Industry Benchmarks
Proficiency levels vary significantly by industry and role. Technology companies tend to have higher overall proficiency, with median scores in the Competent range. Financial services and consulting firms cluster around the boundary between Foundational and Competent. Healthcare and government organisations tend to have lower median scores but are investing heavily in upskilling.
Setting Meaningful Goals
Rather than aiming for everyone to be an expert, set realistic goals based on role requirements. Not every role needs expert-level AI proficiency. For most knowledge workers, Competent is the appropriate target. For AI-intensive roles, Proficient or Expert may be required. The key is having a clear, measurable framework that lets you set goals, track progress, and celebrate improvement.