BCG's AI Fluency Ladder: Which Level Are You Actually At?
Most people asked to rate their own "AI skills" default to a vague number out of ten. Boston Consulting Group's AI Fluency Ladder replaces that guess with four concrete levels, and most jobs only require the first two.
BCG built the ladder around its own roughly 30,000-person workforce, and it defines each level by what someone can actually do, not how confident they feel. L1, AI Explorer, is foundational literacy: understanding what generative AI is and when it genuinely helps versus when doing something manually is just faster, using pre-approved tools and skills the organization already vetted, sense-checking every output before acting on it, and knowing the organization's AI usage policy well enough to stay inside it. BCG treats this level as close to universal -- reported at roughly 100% of its workforce.
L2, AI Practitioner, is where AI use becomes a real part of getting work done rather than an occasional experiment: recognizing when a task calls for a single prompt versus a multi-step workflow versus a proper agent, configuring a reusable AI workspace with the right context and sources for recurring tasks, and quality-checking output specifically for bias and low value rather than just accepting a plausible-sounding answer. BCG reports this as mandatory for the large majority of its workforce -- 65%+ -- which lines up with how much of typical office work now touches an AI tool somewhere in the process.
L3, AI Automator, and L4, AI Multiplier, are where it shifts from using AI well to building with it. An Automator designs and connects workflow steps using approved tools, documents them so a teammate could pick them up and run them, tests before sharing, and makes deliberate calls about what's actually worth automating rather than automating for its own sake -- BCG frames this as optional, reported at 35%+ of its workforce. A Multiplier goes further still: building AI tools other people rely on, validating AI-generated code and output against real quality and risk checklists before handoff, and treating a production deployment as a formal, gated process rather than something that ships when it looks done. BCG reports this top tier at a selective 10-12% of its workforce -- a genuine specialization, not a baseline expectation.
The governance dimension runs through every level and gets more serious as you climb: an Explorer just needs to know what data is allowed into an AI tool before adding it; a Multiplier is expected to trigger a formal risk review before sensitive data ever enters a system they've built. That escalation is the real point of the ladder -- more capability comes with more responsibility for what the system does when it's wrong, not just credit for what it does when it's right.
Based on Boston Consulting Group's published AI Fluency Ladder framework, developed and reportedly deployed across BCG's own workforce.
Get your own AI exposure score
Two minutes, free — see how exposed your specific role is and what to do about it.
Get your AI Vulnerability Score