Licenses purchased. Pilots executed. Training completed. And yet, operations look the same. It’s the most common situation we see in B2B companies that have been investing in AI for a year or two.

The Deloitte State of AI in the Enterprise report (January 2026) — 3,235 C-suite leaders, 24 countries, six industries — puts hard data behind that frustration. And the data confirms what we see in projects: the bottleneck is no longer technology. It’s activation.

60% Have Access. Fewer Than Half Use It

AI tool access grew 50% in a single year. It went from under 40% to around 60% of workers with company-sanctioned tools. A significant jump. But among those who now have access, fewer than 60% use them in their daily work. It’s the same pattern as last year: more licenses, same actual usage.

And when you look at what happens with pilots, the picture gets worse. Only 25% of surveyed companies have moved more than 40% of their AI experiments to production. 54% say they’ll get there “within the next 3 to 6 months” — the same promise from last quarter.

Deloitte classifies companies into three levels based on how much they use AI to change their business:

Level	% of companies	What they do
Deep transformation	34%	Create new products, reinvent processes, change business models
Process redesign	30%	Redesign key workflows around AI, but keep the business model
Surface-level use	37%	Use AI with no change to existing processes or operations

Buying AI licenses and actually activating AI to change results are very different things. The 37% of companies using AI superficially are paying for licenses to do exactly what they were doing before — just with a more sophisticated tool on the desk.

Nobody Redesigned a Single Job

This is perhaps the most revealing data point in the report: 84% of surveyed companies have not redesigned jobs around AI capabilities. Eighty-four percent.

It’s not that they’re doing nothing. 53% are investing in generic training — what Deloitte calls “AI literacy.” General courses on what a language model is, how to write prompts, basic concepts. But only 33% are redesigning career paths and 30% are assessing changes to skills demand.

Meanwhile, automation expectations are accelerating. 36% of companies expect at least 10% of their jobs to be fully automated within a year. 82% expect the same within three years. But if nobody is restructuring the remaining roles, the company is heading toward a scenario where jobs change but job descriptions, approval workflows, and career paths are still from 2023.

This doesn’t get solved with a prompting workshop. It’s a structural problem. If your company has 200 employees and expects 10% of roles to be automated within a year, you need to define what those 20 people will do and how the work AI absorbs gets redistributed. The AI fluency gap requires something deeper than courses: redesigning how work functions from the job description up.

In our experience at IQ Source, resistance doesn’t come from employees. It comes from organizational inertia. HR departments don’t know how to rewrite job descriptions to include AI collaboration. Managers don’t have metrics to evaluate human-machine productivity. And leadership still thinks of AI as a tool you add to the current workflow, not something that should redefine it.

74% Want Agents, 21% Know How to Govern Them

AI agents — systems that reason, use tools, and act autonomously — are entering companies faster than any recent technology. According to Deloitte, 23% of companies already use them at least moderately. Within two years, that number will reach 74%.

But only 21% have a mature governance model for autonomous agents. This is the largest control gap in the report.

The risks companies are most concerned about:

• Data privacy and security: 73%
• Legal, IP, and regulatory compliance: 50%
• Governance capabilities and oversight: 46%
• Model quality, consistency, and explainability: 46%

This isn’t paranoia. Agents do things previous models didn’t: they make decisions, execute actions, modify systems. An agent that approves an expense, sends a client email, or modifies CRM data requires different oversight than a chatbot answering questions.

And there’s a geopolitical factor. 77% of companies now consider the country of origin of AI as a factor in purchasing decisions. Technology sovereignty is no longer just a government issue — it’s now a vendor selection criterion for private companies. If your company plans to scale AI agents, governance needs to exist before deployment, not after.

Productivity Goes Up, Revenue Doesn’t

The Deloitte report shows a paradox that can’t be ignored. When companies are asked what benefits they achieve today with AI versus what they hope to achieve, the largest gap is in revenue.

Benefit	Achieving today	Hope to achieve
Efficiency and productivity	66%	60%
Better data-driven decisions	53%	61%
Cost reduction	40%	65%
Better client relationships	38%	60%
Better products and innovation	38%	60%
Revenue increase	20%	74%

66% of companies already achieve efficiency gains. But only 20% generate additional revenue from AI — while 74% hope to. That 54-point gap won’t close by doing the same things faster.

And there’s another warning sign in preparedness. Compared to last year, perceived preparedness for strategy rose 3 points and for governance rose 6. But for technical infrastructure it dropped 4 points and for talent it dropped 2. Companies feel more ready at the executive level, but less prepared in the operational foundations they need to execute.

This confirms something we see repeatedly: doing the same things faster with AI generates efficiency, not transformation. If your sales team uses AI to generate emails faster but the lead qualification process is still manual, you’re gaining speed without changing the outcome. AI vendor selection should start with revenue impact, not ease of implementation.

What This Means for Your Company

Deloitte’s data isn’t abstract. It’s a mirror. And the question you should ask isn’t “do we have AI?” but “what changed since we bought it?”

Three concrete actions that follow from the report:

Measure pilot-to-production conversion. If you don’t know what percentage of your AI experiments reached production, you don’t have a strategy — you have an innovation budget with no accountability. The global average is that only 25% of companies have moved more than 40% of pilots to production. Where is your company on that curve? At IQ Source, we start every assessment by measuring exactly this: how many pilots you have, how many are in production, and what’s blocking them.

Redesign 2-3 roles in a pilot department. Don’t try to transform the entire company. Pick one department — operations, finance, support — and redefine how 2 or 3 roles function when AI absorbs repetitive tasks. What does the financial analyst do when AI generates the reports? What does the support agent do when AI handles first-level tickets? The answer can’t be “the same as before.” We help map these roles and define the new human-AI responsibilities for each position.

Build agent governance before scaling. If 74% of companies plan to deploy agents within two years but only 21% have mature governance, the opportunity is in being part of that 21% before regulation forces you. Define which decisions an agent can make without human approval. Then implement audits for autonomous actions and maintain a centralized registry of what agents operate across your company. We implement these governance frameworks as part of our AI automation projects.

---

If 84% of companies haven’t redesigned a single job, and yours is in that 84%, the AI licenses you’re paying for are producing less than half their potential value. It’s not a technology problem — it’s an organizational design problem. And it has a solution, but it requires moving the conversation from “what tool do we buy” to “what do we change about how we work.” If that’s the conversation you need to have, we can help you give it structure.