An AI Agent Joins Your Team — Who Manages It?
Quick answer
Before switching on AI agents, assign each one a named owner, a human manager, and clear rules for what it may touch. UK SMEs should sort data access and accountability first — not after something goes wrong.
Before switching on AI agents, assign each one a named owner, a human manager, and clear rules for what it may touch. UK SMEs should sort data access and accountability first — not after something goes wrong.
I see this all the time with UK SMEs. Someone switches on an AI agent — maybe a Power Automate flow that drafts purchase orders, or a chatbot answering customer queries — and six weeks later nobody can tell you who owns it, what it's allowed to touch, or what happens when it gets something wrong.
It's like hiring a new starter and forgetting to give them a line manager. Utter chaos, quietly building in the background.
What the research says
McKinsey's recent piece on HR's transformative role in an agentic future makes a point that's dead simple but almost everyone skips: AI agents aren't just another bit of software you buy and forget. They do work. And anything that does work needs a role in your organisation — an owner, a manager, boundaries, and a way of learning.
Pair that with the BCG / MIT Sloan research showing agentic AI adoption has hit 35% with another 44% planning to deploy, and the message is clear: the firms winning aren't the ones with the fanciest agents. They're the ones who've thought about how the work gets structured around human-plus-agent teams, who decides what, and who's accountable when it goes sideways.
Competitive advantage won't come from early access to AI agents — everyone will have them. It comes from the organisational design around them.
Why it matters for your business
In an SME there's nowhere to hide. If an agent starts sending emails with wrong pricing, or updating ERP records with dodgy data, that's your reputation and your audit trail. And because nobody was named as its owner, the finger-pointing when it goes wrong is a proper nightmare.
I've seen it dozens of times with ERP go-lives — the tech works, but nobody owned the process. Agents just make that failure faster and cheaper to trigger.
Three practical things to do
- Give every agent a manager. One named human, written down. Same as any team member. If you can't name the manager, don't switch it on.
- Define its job description. What data can it read? What can it change? What must it escalate to a human? Keep it tight to start — expand later.
- Review it like a probation period. Weekly check of what the agent actually did versus what you expected. Catch the drift before it becomes a data quality problem.
None of this needs a big programme or a consultant deck. It needs half an hour and a bit of honesty about where your data's fit for an agent to touch in the first place.
Sort the ownership before the agent starts doing real work. Everything else is just tidying up after the mess.
References & Further Reading
- McKinsey, HR's Transformative Role in an Agentic Future
- BCG & MIT Sloan Management Review, The Emerging Agentic Enterprise
- Anthropic, Building Effective AI Agents
Thinking about agents?
I help UK SMEs get their data and ownership straight before letting AI agents anywhere near live systems. Proper foundations, no flannel.
Let's have a chatMatty 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.
Start with a 30-minute data risk call