Once, these were framed as edge cases — something to manage around the margins. That approach is starting to feel outdated.

Complex borrowers showing up in bridging, commercial and specialist residential pipelines today is increasingly the norm, not the exception. Contractors drawing from limited companies. Portfolio landlords with interleaved ownership structures. Business owners whose P&L tells one story while their bank statements tell another.

These aren't difficult borrowers. They're just borrowers whose financial lives don't map neatly onto legacy assessment models.

The question I'd encourage lenders to sit with isn't 'how do we accommodate these cases?' It's 'why are we still treating them as exceptions?'

An infrastructure problem

Most of the friction I see has nothing to do with skill problems. The underwriters I speak to are experienced, capable and well-versed in complex cases.

The friction is almost always infrastructural. Data is siloed. Documents arrive in PDFs that need to be manually decoded. Two underwriters reviewing the same case from different angles end up reaching different conclusions — not because one is wrong, but because the information isn't presented consistently.

In a bridging context in particular, this is enormously important. Speed is often part of the product proposition. But when an underwriter must manually reconstruct the finances of a self-employed developer with multiple live projects — pulling together historical accounts, current cash flow and an exit strategy from separate, unconnected sources — the clock is working against them before they've even begun.

That's not a people problem. It's a data architecture problem.

•   The Finance Professional Show 2025: The Video
•   finova Broker Integrates with UnderwriteMe
•   finova joins IMLA as associate member

Introducing agentic AI

This is where I think the conversation needs to move. The industry talks a lot about AI in lending, but often in fairly passive terms: dashboards, analytics overlays, document OCR. What's becoming genuinely interesting is the agentic layer. This is AI that doesn't just surface data, but actively works through it.

That means pulling verified income data via open banking, structuring unstructured documents and surfacing income patterns across multiple streams without an underwriter having to join the dots manually.

We’re not trying to remove human judgement. In specialist lending, that judgement is irreplaceable and a key part of the value proposition. The goal is to get the right information in front of the right person, already structured and contextualised, so that their expertise can be applied to the decision rather than the data retrieval.

When you frame it that way, it moves from being a technology conversation to become an underwriter empowerment one.

Find your source of friction

I don't think there's a single lender in the specialist space that believes their current tech stack is future-proof. But there's a tendency to wait — for the right integration window, for the market to settle, for a vendor to arrive with a fully packaged answer.

That wait has a cost. It's measured in cases that take longer than they should, in inconsistency that erodes confidence, and in borrowers who find a route elsewhere.

The lenders I find most interesting to talk to are not necessarily the biggest or the most tech-forward. They’re the ones asking the right questions. Not 'what AI tool should we buy?' but 'where in our process does complexity create the most friction, and what would it mean to remove it?'

Start there. The technology exists to help. The case for change, in my view, has already been made.