Human in the Loop: Risk Management and the Use of AI

AI is getting a lot of attention across the industry, but for many mortgage brokers, the question isn’t what it can do, it’s where it fits into your day-to-day work, and how to use it without creating more risk or extra complexity.

In many cases, it’s common to see firms hold back, not because they don’t see the potential, but because they’re unsure how to start in a way that feels safe and manageable.

Start with the parts of your day that slow you down

Most broker businesses already have clear pressure points. It might be preparing for client calls, keeping notes consistent, or staying on top of case follow-ups. These aren’t new challenges, and they’re often accepted as part of running a busy brokerage.

For example, you might use it to:

• Draft a follow-up email after a client conversation
• Turn rough notes into a clearer case summary
• Pull together key points ahead of a review

In each case, the process stays the same. You’re still reviewing, sense-checking and owning the output. It just takes less time to get there. That can mean quicker responses, clearer communication and a more consistent experience for your clients.

This is where AI can start to support your workflow in a practical way.

If you want to get more from AI day to day, this blog on How to use advanced AI prompts to uplevel your work is a useful next step.

You don’t need new systems to see value

One of the biggest misconceptions is that AI only becomes useful once it’s fully built into your CRM or systems.

In reality, many brokers start much earlier than that.

In many cases, it’s about:

• Using AI alongside your existing workflow
• Supporting individual tasks rather than whole processes
• Testing where it saves time without affecting quality

For example, you might draft a client email using AI, then personalise it before sending. Or summarise a case before updating your CRM.

This approach keeps things simple. It also allows you to build confidence before making bigger changes.

Why simple governance helps you move faster

It’s common to think governance will slow things down. In practice, it often does the opposite. The firms seeing the most value from AI are usually the ones that set a few clear guardrails early on.

That doesn’t need to be complex. It can be as simple as agreeing:

• What type of client information should never be entered into AI tools
• Where outputs should always be reviewed before being used
• How AI fits into your existing compliance process

For example, when drafting a client email, you might remove any identifiable details before using AI, then review the wording before sending.

This gives you confidence in how the tool is being used. It also reduces hesitation across the team, which is often what slows adoption.

This is sometimes referred to as keeping a 'human in the loop', a principle the FCA has highlighted as central to responsible AI use in financial services. In practice, it simply means the adviser remains in control of the final output. That's not a limitation. It's what makes AI safe to use in a regulated business.

If you want to understand more about the FCA's position, read our blog. Navigating AI: A broker’s guide to understanding the FCA’s view.

Avoid trying to change everything at once

AI is often positioned as a full transformation. New tools, new systems, new ways of working. For most brokerages, that’s not how it plays out in reality.

It’s more common to see gradual adoption, where small changes build over time.

That might include:

• Starting with internal use before anything client-facing
• Using AI to support preparation rather than decision-making
• Keeping final judgement firmly with the adviser

For example, you might use AI to prepare notes before a remortgage review, but still lead the conversation yourself and make all recommendations.

This keeps control where it needs to be, while still reducing admin.

What this could look like in a typical case journey

In practice, these changes are often small but noticeable.

A typical case might involve:

• Drafting an initial follow-up email after the first conversation
• Summarising key points before progressing the case
• Preparing for a review with a clearer view of previous interactions

Over time, this can lead to:

• More consistent records
• Smoother case progression
• Fewer gaps in communication

It also helps reduce the risk of missing important details, particularly where information might otherwise sit across notes, emails or different systems.

Where this connects back to your CRM

As these habits build, they naturally start to improve how your CRM is used.

For example:

• Notes become more structured and easier to revisit
• Case histories are clearer and more complete
• Information is easier to pick up and use in future conversations

Even without full AI integration, this makes your CRM more useful day-to-day.

If you're thinking about what tools to add next, this guide on The Small Broker's Guide to Building a Smart Tech Stack covers the practical options well.

A simple next step 

If you’re considering where AI fits in your business, it often helps to keep things simple.

You could start by:

• Identifying one task that regularly slows you down
• Testing how AI could support that task
• Setting clear boundaries around how it’s used
• Reviewing outputs before using them in client communication

This keeps the focus on small, manageable improvements rather than large-scale change.

Read more

You can find a growing collection of resources on AI for brokers on our AI topic page, which we keep updated as things move quickly in this space.