HRPulsar
01METHODOLOGY

The Three Levels of AI Fluency.

A practical framework based on Reid Hoffman's methodology — fluency as observable behaviour, not adjectives.

Behind every AI Fluency Score in HRPulsar sits the same scale: ten behavioural markers, grouped into three levels, that describe how individuals and teams actually use AI today. This page is the framework itself — the methodology, the markers, and a 10-question self-assessment.

Take the free assessmentRead the framework →
02ORIGIN

What is this framework and who is Reid Hoffman?

Co-founder of LinkedIn, partner at Greylock, author of The Start-up of You and Impromptu. Hoffman introduced the Three Levels framework as a practical way to describe how individuals and teams adopt AI — from voice-driven chat assistants (Basic) to multi-agent orchestration (Advanced). It is deliberately behavioural: each level is defined by what people do, not by how confident they feel.

“There will no longer be solo specialists working alone. Each of us will operate together with our own set of AI agents.”
Reid Hoffman, LinkedIn co-founder, February 2026
Take the free assessment
03THE SCALE

No vibes. Just behaviour.

Each level groups three or four behavioural markers. The markers are observable: for example, a colleague could watch you work for an hour and tell which level you operate at. The same markers feed the team-level AI Maturity Index inside HRPulsar — your individual score is built on the same scale your future employer will measure.

Basic

0–3

  • Voice mode for chat assistants

    You speak to your assistant on the way to a meeting instead of typing.

  • Role-prompting: assigning a persona to the AI

    You ask Claude or ChatGPT to "act as a senior recruiter" and review a job description.

  • AI-first web research instead of search

    You skip Google entirely and start research with an AI chatbot, then verify the few facts that matter.

  • Iterative refinement: when the first answer is off, refine the prompt and re-run

    You see a vague answer, rewrite the prompt with more context, and re-run instead of starting over.

Middle

4–7

  • Persistent agents (Claude Projects, custom GPTs, scoped Assistants)

    You have a Claude Project named "Weekly leadership update" that already knows your team, format, and tone.

  • AI embedded in recurring work — reports, screening, briefs

    Your candidate-screening pipeline runs through a custom GPT every Monday.

  • AI grounded in your own data sources (uploaded files, knowledge base, database)

    You connect AI to your internal knowledge base instead of pasting context every time.

Advanced

8–10

  • Multi-agent orchestration with hand-offs

    You run a two-agent setup: one drafts, one reviews against an explicit checklist.

  • Cross-run pattern analysis — agents reviewing agents

    You log AI runs to find patterns across many outcomes, not just judge the next answer.

  • Live external integrations, not just chat context

    Your AI workflows pull from APIs and live integrations, not paste-buffers.

Rate yourself now →
04FREE ASSESSMENT

Where are you on the scale? Rate in 2 minutes.

Self-assessment based on behavioural markers from the Hoffman model. Computed entirely in your browser — no account, no save. You'll get a level, a 0–10 score, and three concrete next moves.

AI Fluency assessment · 10 questions

0/10

Foundations

Day-to-day AI use — chats, voice, role-prompts.

01. I use voice mode to talk to AI assistants instead of typing for everyday questions.
02. I start research with an AI chatbot (ChatGPT, Claude, Gemini) instead of a traditional search engine.
03. I write prompts that assign a specific role to the AI ("Act as a senior recruiter…").
04. When the first answer is off, I refine the prompt and re-run it instead of giving up.

Integration

AI embedded in your recurring work, with your data.

05. I use AI agents with a fixed role and persistent context (Claude Projects, custom GPTs, scoped Assistants) — not just one-off chats.
06. AI is embedded in at least one of my recurring work processes (weekly reports, screening, briefs), not only ad-hoc tasks.
07. I connect AI to my own data sources — uploaded files, a knowledge base, or a database — instead of relying only on chat context.

Orchestration

Multi-agent setups, meta-level analysis, live integrations.

08. I orchestrate multiple AI agents that hand off work to each other, or have one agent review another agent's output.
09. I look for patterns across many AI runs or across multiple agents to improve the system, not just the next answer.
10. I pull external data (APIs, live integrations, structured sources) into my AI workflows rather than working only with what's in the chat.

Computed in your browser. We log an anonymous completion only — no answers, no email.

05IN PRODUCT

Apply the scale to your entire team.

HRPulsar runs the Hoffman framework continuously across every employee profile. The result is an AI Fluency Score for individuals, teams, departments — and the company as a whole. Compliance teams see where AI is used without supervision. Heads of People see who scaled past Middle and who needs a nudge. Every score breaks down to the same ten behavioural markers you just answered for yourself.

See the cloud features →
06SHARE

Share your level. Reserve a verified profile.

Self-assessment is the starting point. A verified, portable AI Fluency profile — backed by external review — is what makes fluency credible across employers.

Reserve a verified profileRead the cloud features →
07FOR TEAMS

Get your team's collective AI Fluency Score.

HRPulsar Cloud gives you the team view, the trend lines, and the supervision-gap report — same methodology, all employees.

Start freeSee pricing →