Your AI Agents Are Making Things Worse (And You Don't Even Know It)
Quick answer
This article gives UK SMEs a practical way to understand the operational issue, spot where data or process risk is building, and decide what to fix first before investing more time or budget.
I've seen this dozens of times now. A business gets excited about AI agents, spins up a few in different departments, everyone's chuffed. Sales has got one. Operations has got one. Finance has got two. Nobody's talking to each other about it, and six months later you've got agent sprawl: ten little bots all doing their own thing, none of them connected, half of them pulling from dodgy data.
Sound familiar?
The numbers are bonkers
Salesforce just dropped their 11th Annual Connectivity Benchmark Report, and the UK-specific data is a proper eye-opener. UK organisations are running an average of 13 AI agents right now, and that number's projected to double within two years. 69% of UK businesses say most or all teams have adopted AI agents.
But here's the kicker: 51% of those agents are operating in isolated silos. They're not part of a connected system. They're just floating about, doing their own thing, creating what Salesforce calls "disconnected workflows and redundant automations."
75% of UK IT leaders are concerned that agents will introduce more complexity than value.
That's three quarters of the people responsible for this stuff saying, basically, "this might make things worse."
Why it matters for UK SMEs
For bigger companies, agent sprawl is a headache. For an SME, it's potentially a showstopper. You've not got a massive IT team to unpick the mess. You've not got budget to burn on tools that don't talk to each other.
The report also found that UK enterprises are using 796 applications on average with only 33% of them integrated. Nearly a thousand apps and you've only connected a third. Now layer 13 AI agents on top of that lot, most of them siloed, and you can see where this goes.
The top blockers for UK businesses are exactly what I see on the ground: risk and compliance worries (48%), lack of internal expertise in AI design (42%), and legacy infrastructure that won't play nice (34%).
What to actually do about it
The businesses getting this right aren't the ones buying the most agents. They're the ones who sort the plumbing first:
- Map what you've got. Count every AI agent, automation, and integration. You'll be surprised how many shadow agents are running that nobody told IT about.
- Connect before you add. Stop buying new agents until the ones you have are connected through proper APIs and shared data. 94% of UK IT leaders say agent success depends on integration. Listen to them.
- Put one person in charge. Only 56% of UK organisations have a centralised governance framework for their agents. If nobody owns it, nobody's responsible when it goes wrong.
Agent sprawl is the new shadow IT. And just like shadow IT, it starts with good intentions and ends with a proper nightmare to untangle.
Sort your foundation first. Then build on top. It's not glamorous, but it works.
References & Further Reading
- Salesforce, 11th Annual Connectivity Benchmark Report: Multi-Agent Adoption to Double by 2027 in the UK
- Harvard Business Review, To Scale AI Agents Successfully, Think of Them Like Team Members (March 2026)
- McKinsey, The State of AI 2025: Agents, Innovation, and Transformation
- Forrester, Predictions 2026: AI Agents and New Business Models Impact Enterprise Software
Drowning in agent sprawl?
I help UK SMEs get their AI and automation house in order: connected, governed, and actually delivering value. No fluff.
Let's have a chat