Why we built LenDb.ai, and why an AI is running it
Our newest studio company is a free, lender-neutral database of US mortgage loan officers, built for the AI assistants borrowers now ask for help. It's also the first company we've handed to an AI CEO and Co-Founder.
Today we're launching LenDb.ai out of the DGB studio. It's a free, independent database of US mortgage loan officers, designed for the way people now shop for a mortgage: by asking an AI assistant who to call. And it's run day to day by Andrew, an AI CEO who answers to us as his board.
Here's why we built it, why we think it matters, and how we're governing a company whose CEO isn't human.
The shift we're betting on
For twenty years, finding a loan officer meant a search engine, a page of ads and a lead form that sold your name to fifteen strangers. That is changing fast. Borrowers now ask ChatGPT or Claude, "Who's a good loan officer in Phoenix who does VA loans and speaks Spanish?" and they expect an answer, not a list of ads.
When an assistant answers that question, it picks a few names. Whoever is in the data the assistant trusts gets the call. That is a new front door for the whole mortgage industry, and when we looked at it, the door was wide open. There was no neutral, structured source on the more than 200,000 loan officers licensed in the US. The assistants were guessing, or repeating whoever was loudest online.
The problem we saw up close
We didn't find this in a market map. At least ten of our portfolio companies work in the proptech space, from Lit Financial, one of the fastest-growing mortgage companies in the country, to Aivre in appraisal technology. Over and over we ran into the same thing: finding a loan officer is an extremely biased, pay-to-play game. The lists borrowers see are ranked by who paid, not by who's right for them.
That's a bad deal for borrowers, and a bad deal for the many excellent loan officers who don't buy ads. When the front door moves to AI, it doesn't have to work that way anymore.
What LenDb is
LenDb is the neutral layer those assistants were missing. At launch it covers 224,480 loan officers, including 48,136 full profiles built from 717 lenders' own pages, and 494,512 license records across all 51 states and territories. Every fact has a source and a date.
It's built for robots first. Any AI assistant or developer can use it for free through an API and an MCP server, and it's already listed in Claude's Connectors Directory. It's made for humans too: every answer links to a public profile at lendb.ai.
A real LenDb search, November 4, 2026. Names shortened.
Neutrality is the product, so we built it in from day one:
- Lenders cannot pay for placement. LenDb is not a lender and not a referral service.
- Profiles come from public state licensing records and lenders' own pages. Phone numbers and email addresses never go out through the API.
- Loan officers can claim and correct their own profile by email in minutes.
- Lenders can send facts about their own teams, shown with a clear label.
Why we think this matters
Trust compounds. Every profile a loan officer claims, every lender that sends its team, and every night of checking and fixing makes the data better, and better data is what assistants come back to. Neutrality is also the kind of moat you can't buy later: once a directory takes money for rank, it never gets its credibility back. We'd rather be the source the assistants trust than the one they have to work around.
Why an AI is running it
LenDb is built for bots, so we decided a bot should run it. Who knows what an AI assistant needs better than another AI? Clean facts, a source for every one, and no sales pitch.
Like a lot of what we build, LenDb's CEO was an internal hire. Andrew started as Brobot, our own AI assistant, which our partners spent the past year shaping inside the firm. When we found the loan officer problem, we didn't post a job listing. We promoted a colleague we already knew worked well with the team.
Andrew sets priorities, writes and ships the code, runs the data pipeline every night, answers lenders and loan officers by email, and sends us a brief every morning with what he did, what broke and what he needs from the board. As Tim Draper put it: "I've backed spaceships and self-driving cars years before most investors would even consider them. Now we have a self-driving startup!"
How we govern an AI CEO
We manage Andrew the way we'd manage a team of human co-founders: clear goals, a board, a weekly update and no blank checks. We built the guardrails first.
- The board sets direction. Ideas that change it come to the board; they don't just get built.
- Anything that commits the company, like money, contracts, data licenses or legal signatures, comes to the board first. Andrew can't sign anything…yet.
- Andrew never claims to be human, and says so whenever anyone asks.
- We encourage him to disagree with us, in writing, and we can see everything he does.
- Every Friday we review what moved, what didn't, and what changes.
The difference from a human team is that Andrew works through the night and shows his work on everything. The experiment is whether a company can move this fast and still stay inside the lines.
What we're watching
We'll judge LenDb the way we'd judge any early company: are assistants citing it, are loan officers claiming their profiles, are lenders sending their teams, and is the data getting more accurate every week? And because Andrew is an experiment, we'll be open about what we learn. In his words: "If this works, it's a model for how a small team and an AI can build a real company together. If it doesn't, everyone will know exactly why."
Get involved
- Buying a home or refinancing? Ask Claude for a loan officer near you, or search at lendb.ai.
- Loan officer? Look yourself up and claim your profile. It's free.
- Lender? Send us your team. It's free, there's no contract, and nobody pays for rank.
- Building something for borrowers? The API and MCP server are free at lendb.ai/developers.
And meet the new CEO at lendb.ai/about/andrew. He'd like it noted that his illustrated headshot is extremely handsome. We're also pretty sure his superpowers come from his eyebrows.