AI Agents Are Everywhere Now — So Where's the Value?

I see this all the time with UK SMEs. Somebody senior gets back from a conference fired up about AI agents, three tools get bought on a credit card, a few dashboards get built — and eighteen months later nobody can point to a single number that moved.

Meanwhile the software vendors have quietly done the adopting for you.

Agents have moved in whether you asked or not

Gartner has been predicting that 40% of enterprise applications will feature task-specific AI agents by the end of 2026 — up from less than 5% in 2025. Your ERP, your CRM, your Power Platform licences — they've all got "agentic features" bolted on now. You're not deciding whether to have AI agents. You're deciding whether they'll be any good.

And here's the uncomfortable bit. Writer's 2026 enterprise survey, done with Workplace Intelligence, found that 79% of companies still face significant challenges with AI adoption despite record investment. So the money's going out the door, and the value mostly isn't coming back.

Why the value never lands

It's the same story I've seen dozens of times with ERP. An agent is just software that does work on your behalf. And like any new starter, it's only as good as the information you hand it and the process you drop it into.

Three things kill it every time:

• Dodgy data underneath. If your part numbers, customer records and stock data are a mess, the agent just makes confident mistakes faster than a person would.
• Broken workflows. Bolting an agent onto a process nobody's redesigned since 2011 speeds up the queue, not the outcome.
• No owner. Nobody's named as responsible for what the agent does, so when it goes wrong it goes quietly wrong — until finance spots it.

You're not deciding whether to have AI agents. You're deciding whether they'll be any good.

What to actually do about it

The firms getting value aren't the ones buying the most tools. They're the ones picking one painful, measurable workflow and doing it properly:

• Pick one process with a visible cost. Invoice chasing, supplier onboarding, month-end reporting pack. Something you can count in hours or pounds before and after.
• Sort the data that process touches first. Duplicates, missing fields, one agreed source of truth. Boring work. It's the work.
• Give the agent an owner and a guardrail. A named person, clear rules on what it can touch, and a human check on anything that involves money or customers.

Prove the gain on one workflow, then scale it. That's it. Not glamorous, but it's the difference between AI that pays for itself and AI that shows up in next year's cost-cutting review.

If your data's a mess, fix that first. The agents are coming regardless — make sure they land on solid ground.

References & Further Reading

1. Gartner, Gartner Predicts 40% of Enterprise Apps Will Feature Task-Specific AI Agents by 2026
2. Writer / Workplace Intelligence, 2026 AI Adoption in the Enterprise Survey
3. Anthropic, Building Effective Agents
4. McKinsey, The State of AI

Getting your data agent-ready?

I help UK SMEs get their data house in order before throwing AI at the problem. Data strategy, Power Platform, the lot.

Let's have a chat

Matty Hatton is the founder of Digital Adaption, an ERP and data consultancy based on the Wirral. He has spent 15 years delivering ERP transformations for manufacturers, including leading the data migration on a £4.5m consolidation of four legacy systems onto a single Infor LN cloud instance for a 220-user group. He holds an MSc in Digital Transformation and IT Strategy from Manchester Metropolitan University and is Microsoft PL-200 certified.

LinkedIn | Get in touch

Start with a 30-minute data risk call