AI can do a growing share of the work. Most of your employees still aren’t using it.

There is a strange contradiction sitting inside almost every company right now.

On one side, AI has never been more capable. On the other, most employees have barely changed how they work. The tools are deployed. The licences are paid for. And the gap between what the technology could do and what people actually do with it keeps widening.

That gap is the problem Forsight exists to close. And the data behind it is worth understanding, because it is more nuanced — and more urgent — than the headlines suggest.

The capability is real, but adoption is uneven

Anthropic’s Economic Index, which analyses millions of real anonymised Claude conversations mapped to the US Department of Labor’s task taxonomy, gives the clearest public picture of how AI is actually being used at work. The direction of travel is clear — and so is the concentration.

Here is the part most people miss: usage is heavily concentrated, and effective adoption is far lower than raw numbers imply. Computer and mathematical tasks alone account for around a third of consumer AI usage and nearly half of enterprise API traffic. Vast stretches of the workforce — the finance analyst, the operations lead, the HR manager, the sales rep — sit well below the frontier, using AI occasionally if at all, even when their company has handed them a licence.

The capability exists. The deployment exists. The behaviour change does not.

Why the gap persists

It would be comforting to believe employees simply need another training module. They don’t. The gap persists for reasons that have nothing to do with course completion:

What people assume	What’s actually happening
They need more training	They don’t know what AI can do for their specific role. Generic “intro to AI” training never tells a procurement manager which of their weekly tasks AI could handle.
They’re resistant to change	They don’t trust the output. Without judgment about when to rely on AI and when to verify, many quietly stop after one bad experience.
The tool isn’t good enough	It doesn’t fit their workflow. A tool in a separate tab, disconnected from how work flows, gets abandoned.
Adoption will happen on its own	Nobody is measuring or coaching them. Adoption treated as automatic simply doesn’t happen.

The result is a workforce that has AI available but isn’t AI-ready — and a company that is paying for capability it isn’t capturing.

Why this is the expensive part

The instinct, when AI tools don’t deliver, is to assume the technology underdelivered. The evidence points somewhere else: the tools mostly work. What’s missing is the layer between the tool and the person — the measurement and the coaching that turns a deployed licence into a changed behaviour.

That missing layer is why so much AI spend produces so little. And it’s why measuring individual AI readiness — not org-level strategy, not course completion, but where each employee actually sits on the curve from aware to fluent — is becoming the foundational metric of enterprise AI.

What closing the gap actually looks like

Closing the gap is not about buying more AI. It’s about answering three questions for every employee:

The question	What a good answer looks like
Where are they now?	A verified readiness score against what their role actually requires — not a generic benchmark.
What specifically is missing?	The exact skills and tool gaps standing between them and fluency.
What do they do next?	A concrete, role-specific plan they can act on this week — not an abstract year-long curriculum.

This is what Forsight does. We measure each employee’s AI readiness, identify their specific gaps, and generate a personalised 90-day growth plan tied to the AI tools their company has already deployed. Not to replace people with AI — to make the people you already have far more valuable with it.

Forsight is an AI workforce readiness platform from EpicureAI Labs. To see how it works, book a demo. Source: Anthropic Economic Index (2025–2026 reports), anthropic.com/research.