Pinned Tweet
Today we announced a $150M Series D at $2.3B, welcoming @RibbitCapital as a new investor and continued participation from @a16z, and others. Within 6 months of bringing ValonOS to market, we closed over $200m in contracted ARR, and ValonOS will power 1 in 6 mortgages in America. If you want to work on hard problems that matter, we're hiring across all teams in NYC, SF, and remotely.
Valon retweeted
.@Valon just raised a $150M Series D at a $2.3B valuation. Within six months of selling software, they signed over $200M in deals.
(And they're hiring!)
How the seven-year-old company got there:
- Mortgage is $13 trillion of consumer debt running on a system built before the internet, and no servicer will trust a new platform. So Valon became one.
- Co-founder Andrew Wang read every federal and state regulation, 18 hours a day for six months, and turned it into code.
- They ran their own servicer on the software until it hit 3x the industry's efficiency, sold that servicer to a larger mortgage company, and now sell the software to everyone else. One of the biggest servicers in the US is moving 4 million loans onto it, nearly 10% of the market.
Valon is hiring deployment strategists in NY and SF. Read more about their open roles: a16zjobs.substack.com/p/valo…
@xlindadu @wangandrewd
a16z's Angela Strange sits down with @Valon's Andrew Wang and Linda Du to unpack what it takes to rebuild the infrastructure underneath a $13 trillion mortgage market that still relies heavily on systems designed before the internet.
Linda and Andrew explain why Valon chose the hardest path: becoming a regulated mortgage servicer, translating decades of federal and state regulation into software, and proving the platform on its own loans before selling it to the industry. That foundation made Valon roughly 3x as efficient as traditional servicing and created the system of record it is now using to bring AI into complex mortgage workflows.
They also discuss what AI makes possible on top of that infrastructure, from agents handling long-tail servicing tasks to voice interfaces and more personalized customer experiences. And they explain why deploying the technology into large regulated enterprises is ultimately as much a change-management challenge as a technical one.
01:02 From Soros to mortgage math
03:00 $13 trillion still on 1960s systems
09:38 Why build the whole stack from scratch
11:47 The 3-5 year licensing gauntlet
15:50 Turning 50 states of regulation into code
18:00 Winning the first customers
22:34 What AI unlocks now
29:47 Everything is servicing
32:52 $200M in deals within six months
YouTube: youtube.com/watch?v=ZBDGuT6E…
@wangandrewd @xlindadu @astrange
Valon retweeted
Yesterday, @xlindadu and I announced Valon’s $150M Series D. Forbes profiled our journey and what we think a world with AI agents looks like for the mortgage industry. Thank you, @zoyahasansoomro, for telling our story!
forbes.com/sites/zoyahasan/2…
Read more from our founders @wangandrewd and
@xlindadu about how we're deploying ValonOS and AI to the mortgage industry: valon.ai/blog/series-d-deplo…
Valon retweeted
@Get_Sphere and @a16z are hosting Ground Truth: a half-day event for senior engineers building at the intersection of AI and regulation.
Regulated spaces are the last layer of AI infrastructure yet to be unlocked.
For good reason: they run on unstructured, contradictory and constantly moving rules. They exist as a reasoning problem, and reasoning is what AI made buildable.
We're bringing together the people shipping in the space for lived insights on two technical topics:
1/ Reliability guarantees in production AI + human-in-the-loop design
2/ Compliance-as-a-moat
With the engineers and technical leaders:
- @Harvey: Abhishek Verma
- @valon: Tamjid Rahman
- @raindrop_ai: Zub
- @Luminai: Vamsi Chitters
- @Forus: Anirudh Jain
- @Databricks: Conor B. Murphy
+ others to be announced
Spots limited, RSVP via the link: groundtruth.getsphere.com/
We use AI aggressively and have also turned off AI access for new (non-engineering) hires. Here’s why: 👇
An AI startup CEO says AI overuse increased costs and left colleagues cleaning up AI slop. Now, most new hires must learn the job without it. bit.ly/4zMa98D
and it's never been a better time to join :). open roles at valon.ai
This is my new favorite mission statement: “saving the world from mainframes, legacy systems and outdated infrastructure” @xlindadu @wangandrewd @valon
We're hiring in NYC & SF :) open roles in valon.ai
Valon retweeted
1/
Two weeks ago, we turned off AI access for every new non-engineering hire at @valon.
That is a strange decision for a company at the frontier of AI adoption. It is also one of the biggest bets we can make on our people.
valon.ai/blog/we-turned-off-…
Valon retweeted
1/ Six months ago our design team pointed Claude at our Figma kit. It choked: 200,000 layers, inconsistent tokens. It gave up, screenshotted the kit, and guessed.
That failure taught us what designing with AI actually is.
valon.ai/blog/we-re-all-desi…
Valon retweeted
Asked what he would build if he started over, Patrick Collison did not name an AI lab.
He said he'd go for "the janky 1990s software" running whole industries nobody in the Valley will touch.
3 examples of this:
1) Vertical SaaS
“Deployed engineer” shouldn’t be a buzzword. At Valon, our teams are actively deploying AI-native software across some of the most complex and regulated environments in finance.
Want to learn more about the work and meet the team? Join us for a round of golf in NYC on the 13th.
Register here: luma.com/mozklm4q
Valon retweeted
pushing to prod on a Friday but at least we have unlimited Claude Code + Codex tokens and a coffee bar @valon
Valon retweeted
Welcome to the world of mortgage servicing!
Asked DeepSeek V4 Pro to make a mortgage servicing themed romhack of Pokemon Red
7 years ago @BrianJMcGrath and @wangandrewd wrote the business plan for @valon on the back of a napkin. Now, he's all in as Executive Chairman.
Here's why: valon.ai/blog/Brian-McGrath-…
Valon retweeted
we're using this in production. examples:
- eng team loves a specific PM's PRDs? extract their taste, generate PRDs that read the same
- AI gets the facts right but the output reads like slop. this fixes slop by anchoring to what "good" actually looks like
- tested it across PRDs, process docs, and other domains. adds quality back in automatically
the key insight: with AI its easy to get correct. the hard part is getting good. this is a system for bottling "good."
autoreason constructs taste from debate. but whose taste? the judges'. that's undirected optimization.
what if you only need one golden doc? anchor to it. extract criteria. hill-climb. when the eval saturates, the gap reveals criteria no human would have thought to specify. extract again. repeat.
debate uses a human-defined jury as the loss function. this RLs the loss function itself.
Valon retweeted
We built a pipeline that mines our PR review threads and turns the recurring patterns into automated review rules
Convention nits from human reviewers dropped ~40% and shifted to design questions.
Write-up here: valon.ai/blog/your-best-engi…
Skill here: github.com/valon-technologie…
