ValonInterview Hub
The thesis · From the CEO

Our strategy is an
open secret.
It’s just hard to copy.

Silicon Valley overvalues growth rate and undervalues durability. There’s a massive spread between what seems impossible and what is merely unattempted. Our question: where does entrenched complexity suppress competition long enough that patient, methodical execution earns an astronomical reward? Mortgage servicing was our first answer. It will not be the last.

Written byAndrew Wang, CEO & co-founder
The setup

A $13T mortgage market running on software older than the internet.

The two dominant platforms in mortgage servicing — MSP (BKFS) and Loanserv (Sagent) — were architected in the 1960s and 1970s. Since 2008, regulation has only multiplied. Servicers adapted with band-aids: short-term fixes on legacy systems, ever-larger operations teams.

Software wedge strategies fail here. Servicers operate on razor-thin margins and won’t pay for incremental features. Anyone trying to build it ran out of capital long before they had a product. So no one tried.

$25B
annual industry spend
Acyclical
payments don’t care about rates
7 yrs
and no modern system of record has emerged behind us
The opening

Complexity is the moat. Conquering it is the innovation.

The casual innovator is deterred by the barriers and the carcasses of past attempts. To us, the overwhelming complexity is music. The only way to win is to take on the whole problem — build the servicer, prove the economics at scale, then sell the platform.

That’s a 2-3 year licensing trek, 50 state approvals, GSE certifications, and an operations company built from scratch — just to earn the right to ship enterprise software. Most teams would call this impossible. We called it the plan.

The proof, in unit economics

Step-function efficiency, not incremental.

The reason this works is the gap between what the industry runs at and what ValonOS runs at. The same loan, serviced on Valon, costs 40% less — and customer satisfaction goes up, not down.

Industry incumbent

Benchmark servicer · legacy stack

~2×
cost per loan vs. ValonOS, illustrative
Ops headcount
High
Tech & vendor fees
Low
Same loan · on ValonOS

Powered by workflow automation & LLM agents

illustrative baseline
Ops headcount
Lower
ValonOS platform
Higher
Approximate margin delta to incumbent~40% less
Illustrative only · directional shape, not contract pricing. Exact figures confidential for competitive reasons.
Four phases · One playbook

The strategy, in four steps.

Each phase is deliberate. We picked the order because each step buys the next. Click through to see where we are and what’s next.

Step 1: Prove the thesis on the hardest version of the problem.

Nobody buys mortgage software until you’ve proven it works at scale. The bar in the industry: 500,000 loans, all major loan types, regulator-grade audits. So we built our own subservicer, took on the operational complexity, and used it to iterate on the platform.

At peak: ~$130B UPB, ~600K mortgages, a top-10 subservicer. First and only fintech with Fannie, Freddie, and Ginnie approval. SOC I & II, USAP, Fannie STAR, Freddie SHARP.

In May 2026, we entered into a strategic partnership with Carrington that includes the acquisition of our subservicer. Phase 1 is complete. The proof is done. The OS remains.
~$130B UPB at peak~600K loansTop-10 subservicer90% CSATFannie · Freddie · Ginnie approved
Trajectory · We said it, we’re doing it

Two years to a different company.

The shape of the next 24 months isn’t a forecast — it’s contracts already signed.

End 2025
~700K
loans live on ValonOS

Phase 1 model · Valon Mortgage subservicer at scale

Done
Today · May 2026
~1M
loans live on ValonOS

First enterprise software customer live (Feb 2026, NDA) · Carrington strategic partnership announced

Live
End 2027
~5.4M
loans live on ValonOS

Newrez 4M loans go live · Carrington 800K onboarded · multiple enterprise contracts on one platform

Booked
Loans live on ValonOS. Sources: Carrington strategic partnership (May 7, 2026); Newrez deployment announcement (Feb 2, 2026, go-live 2027); first enterprise software customer live Feb 2026 (named under NDA). “Booked” indicates contracts signed and onboarding scheduled.

Mortgage was the cost of building the option. Now we exercise it.

We are mapping every regulated services category with the same filter we used for mortgage: hatred for the incumbent, menial work caused by broken systems, cash and accounting at the core, correctness as a defining feature, AI-driven operational leverage available. The list is long. We’ve already done the hardest one.