Why the same AI tool wins for one team and fails for the next.
A question we get on most first calls: "We bought the enterprise licences. Why isn't it working?" Usually the licences are fine. What's missing is fluency — and crucially, shared fluency across the team.
The same model produces excellent results for one team and mediocre noise for another inside the same company. The difference is rarely the technology. It is whether the team has a shared way of deciding what to delegate, describing it well, judging the output, and owning the result.
From usage to understanding
The trap is measuring adoption — seats activated, messages sent — instead of fluency. High usage with low fluency is how you get confident, fast, wrong work at scale. The shift that matters, as Anthropic frames it, is from scaling usage to scaling understanding.
Mapped to the business, the 4Ds become four capabilities:
- Delegation → strategic: where do people, AI, or both own the work?
- Description → operational: can the team brief AI consistently?
- Discernment → analytical: do we catch bad output before it ships?
- Diligence → governance: who is accountable, and is data handled properly?
How to build it without a six-month programme
Three moves, drawn from how Anthropic and its academic partners suggest organisations adopt the framework:
1. Baseline first. Get the team talking about how they already use AI and where they want to go. You cannot improve what you haven't measured. (Our assessment exists for exactly this.)
2. Translate to roles. A support team might prioritise Description and Diligence; a product team Delegation and Discernment. One shared framework, role-specific emphasis.
3. De-stigmatise the conversation. Make actual time for teams to talk about what is and isn't working with AI. Shared language is what makes AI "thinkable and talkable" — the first step to safe experimentation.
Fluency turns AI from a novelty into a management discipline: measurable, repeatable, owned.
That is the whole game. Teams fluent in the 4Ds move faster, decide more clearly, and make fewer expensive mistakes. The rest chase hype without leverage.
Content informed by Anthropic's AI Fluency resources. Further reading: Anthropic — Upskilling orgs with the 4D Framework ↗ · Anthropic — AI Fluency Framework ↗