AI has moved quickly from industry talking point to commercial priority. For mortgage brokers, the question is no longer whether AI matters. It is where to start, how to apply it properly, and how to avoid wasting time on tools that don’t solve the real problem.
Too often, firms begin by looking at technology first. They sit through demos, compare features and talk to vendors before they have properly defined what they actually need AI to do. For brokers, that’s the wrong way round.
The first question should be simple: what problem are we really trying to solve?
A vague ambition to “do something with AI” is not a strategy. The starting point has to be operational. Where are advisers and administrators losing time? Where are cases slowing down? Where are clients being chased repeatedly for missing or incomplete information?
In mortgage broking, the same pattern appears time and again: gather, check, chase. Gather documents, check whether they are complete, chase what is missing, then repeat when something else is not quite right.
That cycle creates real friction. It slows cases down, frustrates clients and takes brokers away from the work that matters most: advising, structuring cases and managing relationships.
The next question is whether the firm is thinking about AI in the right way.
Many brokers still view AI as traditional software: a tool that completes isolated tasks or follows rigid steps. But modern AI, particularly agentic AI, is better understood as though you are bringing a new person into the business.
If you hired a new administrator, you would not simply tell them to press a button. You would explain the outcome you needed, the standards expected, the context of the client journey and when to escalate something to a human. That is a more useful way to think about AI.
The real opportunity is not simply task automation. It’s using AI to complete bounded admin jobs properly and consistently, within the rules and standards of the business.
Another important question is: how mature are we in our understanding of AI?
Some broker firms are still learning. Some have experimented with tools such as ChatGPT. Others are already thinking about how AI could support live operational workflows. The important thing is honesty. Do you understand what AI can realistically do today? Have you moved beyond surface-level experimentation? Have you thought about oversight, governance and client experience?
That leads to another, deeper question: who is accountable for the outcome?
AI can increasingly perform tasks that once required people. It can draft, summarise, analyse, chase, check and progress work. But it can’t currently take accountability for whether the overall process is working properly, whether the right priorities are being set, or whether the results are reliable enough to trust at scale.
That’s an important consideration for those brokers who are taking a DIY approach to AI. A broker may be able to build something useful quickly. But building something flexible, secure, resilient and scalable enough to run operationally is a very different challenge.
Firms need to decide whether they want to own that capability themselves, with the technical, operational and governance responsibilities that come with it, or work with a specialist partner that already carries that expertise.
Brokers should also ask: are we starting in the right place?
The best starting point is often not the most exciting use case. It is usually the admin that slows the business down every day: document collection, missing information, client follow-up and case packaging.
Finally, the commercial case must stack up. How much time is currently being spent on manual admin? How many cases are affected? What improves if that process becomes faster, more accurate and less repetitive?
AI adoption should not be about chasing hype. For brokers, it should be about asking better questions, identifying the right operational problem and freeing up more time for advice.
