A transparent methodology: five literacy vectors, five maturity levels, three layers, and a ranked opportunity list at the end. Here is exactly what we measure and why.
AI readiness is an organization's measured ability to put AI to productive use: the skills of its people, the workflows of its teams, and the governance, data and infrastructure around them. It is not a single score; it is a profile across people and organizational dimensions, because a company can be strong on one and weak on the other.
drivechange.ai measures it with a free, self-serve assessment that takes about 30 minutes per person and produces evidence, not opinions.
AI literacy is behavioural, not self-declared. People overrate their own AI skill, so every scored question asks for a behaviour or a demonstration ('describe the last thing you built'), never a self-rating. The gap between claimed and demonstrated skill is itself a finding.
Each person is scored on five independent vectors, because a single scale hides the truth: two people can score 55 for opposite reasons, one a confident daily user with no risk awareness, the other a cautious expert who rarely bothers.
Vector scores map to five levels adapted from Bloom's revised taxonomy, each a 20-point band. The boundaries that matter most: 2→3, does this person actually use AI with intent?, and 3→4, do they build, not just consume?
Designs AI-enabled processes end to end: evaluates models, reasons about risk and ROI, and sets the direction others follow.
Tools get chained together, light automations get built, and the limits of agents and RAG are understood: AI starts shaping how the work itself is structured.
Effective prompts, the right tool for each task, and regular use with intent. AI is a working part of the day-to-day.
A few tools are known and used occasionally, but ad hoc, with no deliberate technique and no repeatable habits.
AI is equated with ChatGPT and used as a search box, if at all. It has no deliberate place in how work gets done.
Interventions differ at each layer, so the three are reported side by side, never averaged into one number. Training individuals doesn't help when the real blocker is governance, and strategy doesn't help when people can't use the tools.
One person's understanding and behaviour with AI, scored on the five vectors.
Typical failure it exposes: skilled with tools but no technique, or confident misuse.
How a function embeds AI in shared workflows and tooling: aggregated individuals plus team-level questions.
Typical failure it exposes: islands of excellence that don't spread.
Governance, data readiness, infrastructure, funding and strategy: from a leadership diagnostic plus a technical audit.
Typical failure it exposes: strategy with no enablement, or enablement with no strategy.
Your scores are placed against a published-research reference band for your own industry and company size: built from cross-sector studies by sources such as the US Census Bureau, the Federal Reserve, McKinsey, BCG, Deloitte and Cisco. An absolute 55 means nothing without context.
Every comparison names its source and date. Peer cohorts from assessment data appear only once at least 30 organizations of the same industry and size have participated, and individual organizations are never identifiable.
The level structure adapts Bloom's revised taxonomy (remember, understand, apply, analyse, create) and standard technology-adoption maturity models. Every question carries a scoring rubric; open answers are graded against a five-level rubric anchored to observable behaviour, so a human scorer and an automated scorer land in the same band.
The methodology is registered with Safe Creative — Registration #2607186431567.
About 30 minutes per person. Team and organizational views build up as more people complete it.
Yes. The assessment is free and the results are yours. AInvirion offers paid enablement, automation and governance services when you want to act on the results, that part is optional.
Everyone whose work could involve AI, not just engineers. Expectations are role-relative: a level that is under-performing for an engineer can be exactly right for an operations manager.
Sensitive data is encrypted per organization. Benchmarks only ever use anonymized aggregates, and peer cohorts appear only above the 30-organization threshold.
No. Questions ask about everyday behaviour with AI (what you use, how, and for what) in plain language, in English or Spanish.
Start with the gap: where people readiness and organizational readiness diverge is usually the story. The ranked opportunity list tells you what to tackle first.