How might Artificial Intelligence (AI) enhance the mortgage lending process?

21 September 2023

AIAutomationClient ExperienceMeeting of MindsMortgagemortgageRisk

Expert and Facilitator: David Owen, Wealth Proposition Director, Openwork


  1. AI’s ability to enhance the mortgage lending process is a huge opportunity, which could in practice include:
    • Providing real-time, personalised customer experiences, helping mortgage advisers better understand their customers’ needs and preferences
    • Automating routine tasks e.g., data entry and document processing
    • Examining data from various sources to provide insights into market trends and customer behaviours
    • Enhancing risk assessment through AI algorithms that can analyse vast amounts of data to help mortgage companies make better risk assessments and underwriting decisions
    • Optimising processes by looking at data from past transactions and identifying areas for improvement in the mortgage lending process
    • Using machine learning algorithms to detect potential fraud in mortgage applications and prevent illegal activities
    • Automating document management, making it easier for mortgage advisers to store, manage and access customer information and loan documents


  1. When it comes to big data there are 5 Vs to consider:
  • Volume, how much is available
  • Velocity, the speed of data access and transmit
  • Variety - the diversity of it, how accessible
  • Veracity – how reliable it is
  • Value – what it is worth

Discussion points:

Bringing to market new products to help with reliability and efficiency is complex process so having a Chatbot to answer questions and being able to provide the confidence on what is right for the consumer would be hugely valuable.

A big change could be to give the broker everything they need up front, as well as forming partnerships e.g. with an accounting firm that you can source clients who are coming to the end of their fixed rate deals

We currently have 1.9 million paper files in storage. Do we spend the money to digitise these?

With the future picture for Open Banking, the broker model using AI could transform the nature of the business, particularly when it comes to keeping in touch with the clients and looking at the pattern of spending e.g. new build protections.

The future of a mortgage broker is one in which the machines do the heavy lifting and the interpretation, while the planning is then done by the human.

Putting the broker in the position where they don’t have to do the admin, the wrangling that everything stays on track with the application.

The biggest transformation would be in getting the broker to do more obligations to the customer, as only 20% are currently doing the protection/Consume Duty obligations.

Where you’ve got technology, you also have an option to use this all the way through the process, but the broker still wants to pull it out of the system and into the independent offer. So even if you have the technology providing all the systems, if the trust isn’t there then the two systems of broker then lender will continue in order to ensure it will be done.

We are investing heavily in digital solutions and direct-to-consumer digital solutions which has traditionally been seen in the big 6. In the UK, the main challenge with digitalisation is that the app to completion conversion is low, the need to talk to a human is too great.

Product transfers means brokers are not staying in touch with the customers, or keeping them as a lifelong customer – instead they are moving on to the lender, which is not as safe.

Consumer behaviour is changing. If you look at acquisitions, banks’ digital businesses are replacing the branch business, but the dynamic and proportion of intermediaries is still the same. The lenders are also be a threat to brokers in terms of product transfer.

There is a need to modernise our businesses, particularly as younger people will be more inclined to just click on an App and not talk to anyone, this is the direction it’s going.

People want technology but they want to see value. So you have to do the value piece on the loyalty, that they know they are being looked after and someone is there for them if things go wrong.

Lenders are realistic that when you have end-to-end, digital-only that a certain amount of the business will go through them due to the complexity. We are a long way from getting the majority of our customers to get on board with this, or indeed populating the CRM and taking someone all the way to the end.

When it comes to the problem of scaling and deploying AI, in financial services we are too siloed, and with so many small firms it is more difficult.

From an educational point of view, regulation qualifications need to evolve. The AI bot can create questions for the student or broker to think about asking the client - prompting them to think about what they wouldn’t normally, and put this into their process rather than a broker just taking the path of least resistance. 

But what would the regulator feel about AI giving advice? Lenders are all following a scripted process. So the tech is there to do it but what about vulnerable customers etc

So the commercial case is to sell more but this is why mortgages are so behind because the lenders are making loads of money and are not losing sales by not investing in tech. All of the lenders need to go forward 

The model would only work for the big 6

The art of mortgage lending and advice must not be lost - such as reducing the mortgage process to a tick the box exercise.

When it comes to the FCA requiring us to prove why we haven’t gone to the other lenders, evidence of research is our problem.

Ultimately, the technology to sort out the mortgage process has been talked about for decades but we still have the problem with conveyancing and surveying – there is nothing linking this together. AI has a really big job to do and could solve the challenge of bringing the whole process together - but this will only work if it goes across everything.

While there is huge potential if the whole process is fixed - you could get a customer in weeks earlier - but is this needed? How much quicker does a house move need to be?

The mortgage journey is a small part of the value chain, but anything that gives more certainty to any market is a benefit – particularly to the market we are in at the moment with chains falling apart so quickly. Having a slick process to repair the chain more quickly would have the most value.

Key takeaways:

  • The AI is only as good as the data available so have to pick out the small parts rather than the big
  • It's not the wild west - our lenders are privately invested so we can look a little outside the norm
  • Going forward we need the quality assurance of the advice being given - so integrating the AI into the pitch, the tone, the pace - the soft stuff but it can’t take away the quality of advice - the advice point of view is very difficult to AI even if we have a slick and efficient process. This doesn’t yet exist in the corporate world
  • Within the mortgage market AI will only work if it is really taking over the market and would need sufficient investment and data for everything to work along the chain – all the 5 Vs all being met
  • The fundamental point is that the majority of clients wants a human to tell them if they are doing the right thing, to talk it through and we will never get rid of that
  • Given the trust element, this means that AI is not a danger when it comes to human jobs in the financial services industry, as it is in the creative industry, where people don’t need reassuring