A Mortgage Holding Company Says the Industry Digitised the Paperwork and Left the Work Alone

The mortgage industry has spent billions of dollars on digitisation. A borrower can apply online, upload documents to a portal and follow an application through a dashboard. Behind that interface, a great deal of the work required to produce a mortgage is still done by hand.

Ace Watanasuparp, founder and chief executive of Swish Holdings, treats that gap as the sector's defining problem.

"For decades, the mortgage industry has manufactured loans by hand," he said. "Swish is building the technology to manufacture them through software."

What was actually automated

His argument about the past twenty years is narrow and, on the evidence, hard to dispute.

"We've taken paper applications and turned them into online applications," he said. "We've taken physical documents and turned them into PDFs. We've created portals and dashboards."

None of which removes a task. Somebody still reviews the documents. Somebody still moves information between systems, checks it against lending guidelines, clears conditions, prices the loan and manages the file. A single mortgage can pass through a number of separate products belonging to different vendors, and each boundary between them creates an integration to maintain and a handoff where something can be dropped.

The digitisation, in other words, improved the borrower's experience of waiting and left the reason for the wait untouched.

We've taken physical documents and turned them into PDFs

The proposed answer

Swish Holdings is a financial holding company that builds and invests in financial-services businesses, weighted towards mortgages. Its technology arm, Swish Labs, is building what the company describes as an AI-native, vertically integrated mortgage platform.

The distinguishing claim is scope. Rather than selling lenders another component to bolt on, Swish Labs is attempting to own the whole lifecycle through one proprietary stack: customer relationship management, point of sale, loan origination, pricing, AI-powered underwriting, and eventually capital-markets technology.

The thesis is that artificial intelligence can eliminate parts of that work rather than merely digitise them, and the target is everything that happens after the borrower presses submit.

Why vertical integration is the interesting part

The AI claim is the one that will get attention and the integration claim is the one that carries the risk.

A lender running six vendors has six contracts, six roadmaps and six sets of integration debt, and that fragmentation is a real cost. Collapsing it into one stack removes the seams, and if underwriting decisions can genuinely be automated against guidelines rather than merely assisted, the saving compounds across every file.

The difficulty is that owning the entire chain means being correct about all of it. A point-of-sale product that disappoints can be swapped out. A vertically integrated platform that disappoints has to be replaced in full, and lenders know it, which raises the bar for adoption considerably.

There is also the regulatory question that sits over any automated credit decision. Lending is among the most closely supervised activities in the economy, and a model that declines applicants has to be explainable to a regulator in terms that satisfy fair-lending rules. Nothing in the company's framing suggests it is unaware of this. Nothing in it addresses the point directly either.

What would count as evidence

The pitch is coherent and the diagnosis is sound. Manual work behind a modern interface is a fair description of where mortgage lending has ended up, and the observation that the industry automated its paperwork rather than its labour is worth more than most sector commentary.

What is not yet available is the part that matters: production volume, measured cycle times against a lender's previous stack, and the proportion of underwriting decisions that genuinely clear without a human touching them.

Every mortgage technology company of the past two decades has promised to remove the handoffs. The interfaces improved each time. The closing timeline did not move much. Whether this one is different is a question that gets answered by loan files, not by architecture.