MQube mortgages go live on MAB with firms working on AI-enabled fact find – exclusive

MQube mortgages go live on MAB with firms working on AI-enabled fact find – exclusive

 

The firms are also working on an artificial intelligence (AI) system to connect the lending process right from the initial broker fact find.

MQube launched its lending brand alongside the MPowered lender hosting platform earlier this month and is now rolling out product availability to advisers.

This initial buy-to-let product offering is available to limited companies, portfolio and individual landlords and will soon be extended to include houses in multiple occupation (HMO).

MQube distribution director Emma Hollingworth (pictured) told Mortgage Solutions that in addition to the MAB launch, agreements had been reached with several distributors and would soon go live.

“We’ve been running a soft launch with five or six advice firms each from five distributors as we want to make sure everything is well tested and broker ready,” Hollingworth said.

“We went live with TMA last week and MAB now and we have contracts signed with a lot of other distributors, so probably over the next six or seven weeks, all being well, we’ll be adding those too.

“But we want to make sure this is controlled, that it’s a really good experience and we’re getting the right service to brokers,” she added.

 

AI-enabled fact find

At the moment the MPowered Mortgages unregulated buy-to-let products will appear through network and mortgage club sourcing systems and advisers will then need to log in to the MPowered platform to access them and submit applications.

However, Hollingworth revealed the firm has plans to make a straight through process direct from fact find to underwriting, which is being developed as part of an innovation lab partnership with MAB.

“At the moment we are just a standalone system, however our vision ultimately is to connect fact find to underwriter in real time,” Hollingworth continued.

“We’ve formed the artificial intelligence (AI) lab with MAB to explore the opportunities we have to use AI and deep learning in the broker fact find process in the same way we do in the underwriting process.

“That is the plan to work with MAB to develop that process.”

Hollingworth was reluctant to put a timeline on such a development, but noted: “It’s a very agile, fast-paced world. I wouldn’t say this will be long-term, we’re not talking years.”

MAB said it believed implementing AI in the sales process at scale would be the future of the mortgage market.

The broker firm explained that by applying AI at the beginning of the sales process this can reveal customer needs by analysing the information supplied in the mortgage application process, aiding advisers in making appropriate recommendations for products and services, at speed.

Peter Brodnicki, CEO of MAB, added that the two businesses shared an ambition in using AI and deep learning throughout the mortgage process.

“We have a long-term mindset with this partnership and so we’re excited to continue supporting MQube and MPowered as the mortgage application process evolves further,” he said.

 

Data-driven underwriting

Both MQube’s lender hosting platform and it’s own lending offering are based on advanced use of AI, machine learning and rapid access to data.

And the intention is for this to continue with any third-party lenders that join the system along with any extension of the MPowered mortgage products.

“The system doesn’t generally ask anything that we can’t get from data sources or the documents that are uploaded,” Hollingworth said.

“So our platform reads those documents in real time, verifies those documents, reads the data and categorises the data.

“Then the lender’s rules and policy and criteria are applied, so if a lender doesn’t accept someone with a county court judgment (CCJ), because we’re pulling all the data upfront with a soft search, it will find that and tell the broker they can’t go ahead with the case.”

Given this approach it will be vital that brokers have customer documentation ready to upload when submitting the case, rather than waiting for underwriters to request them.

The system can also identify queries and ask the broker for more information directly as the case is being submitted before going to an underwriter.

For example, if there has been an unexpected cash deposit it will ask what this is for and the advisers can reply immediately.

“It’s not just vanilla automated data its quite complicated; we’re reading buy-to-let portfolios, accounts, lots of things, but we’re doing it in real time,” Hollingworth continued.

“That’s presented to the underwriter who will then check the case and it means we can get certainty for the customer with an initial binding offer.”

The funder of this initial £2bn unregulated buy-to-let offering, an unnamed top 15 global bank, requires a physical valuation to confirm the loan.

However, Hollingworth noted that MQube has devised an “automated valuation model on steroids” which uses around 180 bits of property data and provides an early warning to the broker if the values used are not likely to match up.

Once the physical valuation is completed and if everything matches up the offer is formalised.

 

All core areas of lending

Along with the wider distribution for its own products, MQube is in discussions to add more lenders onto the MPowered platform, although confidentiality agreements restrict many details.

But the aim is to have all key parts of the mortgage market covered.

“We are speaking to a lot more lenders and funders and there will be more added to the platform in 2021 and when we are able to we will announce them,” Hollingworth said.

“It’s our intention to be in all the core areas of lending but this will be subject to our FCA permissions, and we’ll announce more about that when we have those permissions,” she added.

The MPowered range of unregulated buy-to-let mortgages includes two and five-year deals up to 75 per cent loan to value (LTV).

Rates start at 2.94 per cent for a two-year fix at up to 50 per cent LTV for individual landlords and all deals have a 1.5 per cent product fee.