The Mortgage Professional’s Guide to Prompt & Context Engineering

Artificial Intelligence is no longer just a concept for our sector; it is beginning to reshape how mortgage professionals operate across the UK.

From brokers and protection advisers to case managers and compliance teams, tools such as ChatGPT, Gemini and Co-Pilot are becoming everyday assistants. They can help draft emails, simplify complex mortgage concepts, create client guides, and even support internal processes.

But the quality of the answer you get from an AI is only as strong as the question you ask. This is why prompt engineering and context engineering are becoming critical skills for anyone looking to maximise the benefits of working with AI.

Contents

• What is Prompt & Context Engineering?

• How do you actually get engineering? 

• Poor & Strong Prompt

• A Key Principle is Iteration

• Watch Out For Hallucinations

• Final Thought

What is Prompt & Context Engineering?

Prompt engineering and context engineering often get mentioned together, and while they’re related, they serve different purposes. Prompt engineering is about how you ask. It’s the craft of structuring your request in a way that directs the AI clearly.

For example, asking it to write a two-paragraph email in a professional yet approachable tone.

Context engineering is about the detail you provide that makes the AI’s response relevant and specific, such as explaining that your client is a first-time buyer with a £25,000 deposit looking at a £200,000 property.

Put simply, the prompt gives direction, and the context provides relevance. You need both to get answers that are accurate, tailored and useful.

How Do You Actually Get Engineering?

Start with context.

Always outline the client’s situation in detail, because this is the single most powerful way to guide the AI towards accuracy. Without it, you’ll only receive generic responses that could apply to anyone.

Define The Task

Once you’ve given the background, define the task.

Be clear about what you want the AI to produce. This could be a step-by-step mortgage guide, a client email, a comparison of limited company versus personal buy-to-let, or a client friendly compliance summary.

Set The Role

After that, set the role by asking the AI to act as an experienced professional aligned with your task, such as an UK mortgage adviser, compliance officer, or marketing consultant. This helps the AI to frame its answers in the right professional voice.

Shape The Tone & Format

Finally, shape the tone and format. Do you need 200 words in plain English? Should it be written as bullet points or as paragraphs? Should it end with a call to action? These instructions are the finishing touches that make the content practical and client ready.

Poor & Strong Prompt

The difference between a poor and a strong prompt is striking.

If you simply ask, “Explain buy-to-let mortgages,” you’ll receive a bland, textbook answer.

If instead you say, “My client is a first-time landlord with a £100,000 deposit and wants to buy a £400,000 property. Please explain the pros and cons of a buy-to-let mortgage for them. Act as a UK mortgage adviser and write in clear, friendly language under 300 words, using a bullet-point structure and finishing with a call-to-action,” the output will be tailored, clear and immediately usable with clients.

A Key Principle is Iteration.

Just as you would never expect a fact-find or suitability report to be perfect on the first draft, you should treat AI in the same way. If the output feels too technical, ask it to simplify. If it misses important details, ask it to expand.

Think of the AI as a junior colleague who is helpful, quick and capable, but always in need of your guidance to reach the right standard.

Watch Out For Hallucinations

Finally, always watch out for hallucinations. AI sometimes generates information that sounds convincing but is not accurate. In mortgages, this could mean misquoting rules, referencing outdated buy-to-let tax changes, or inventing lender criteria that does not exist.

Your role as a professional is to fact-check and ensure compliance before sharing any AI-generated content with clients.

Final Thought

Prompt and context engineering is more than writing better instructions. By providing context, defining the task, setting the role, guiding the tone, refining through iteration, and of course fact-checking, mortgage professionals can

harness AI to save time, improve communication and deliver better client outcomes.

Liz Syms

CEO of Connect for Intermediaries, Chair of the Society of Mortgage Professionals