The Real Competition for Dental Software? Betty, Who Quit Two Weeks Ago

Source: a16z | Published: 2026-07-30T14:30:01Z

a16z's Alex Rampell argues that the biggest competitor to dental practice software isn't another tech company — it's Betty, the office admin who just resigned. One clinic that replaced manual envelope-opening with automation reached 98% billing automation.

Steijn followed him into the back office and couldn't shake what he saw: a Yelp #1 dentist spending 200 hours a month on paperwork. Manually submitting insurance claims. Sitting there waiting to get paid because he couldn't find anyone to handle the billing.

Steijn's first instinct was: surely someone has already solved this. His mother worked at a Dutch hospital in the seventies — back then she'd carry a bag of cash to the bank and process payments by hand. This was America in 2020. It couldn't still be like that.

Then he drove out to Scranton, Pennsylvania, found a gastroenterology practice, and saw the exact same thing. That's when it clicked: this wasn't a Dr. Quan problem. This was the daily reality for 160,000 practices across the country.

They Went In and Did the Books Themselves

A product manager from Robinhood and a product manager from Superhuman walked into a dental practice and asked for a job: let us take over your billing.

Dr. Quan said yes. Dr. Sha in Scranton put them right behind the front desk and told them: "Look at everything. Treat it like you're teaching my son."

This is where Lassie began — co-founders Steijn and Frederick actually sitting there, opening insurance company portals, manually processing claims, reconciling bank statements. When the mail carrier came in and dropped off a stack of envelopes, Steijn opened each one by hand, deposited hundred-thousand-dollar checks into the doctor's bank account one by one, then cross-referenced every line against the itemized statements inside.

They weren't studying the business. They were running it.

That's also why Lassie was able to reach 98% automation: they understood what the work actually looked like better than any engineer could.

Software Never Really Made Work Disappear

a16z partner Alex Rampell has a theory he's shared many times. He laid it out in full once more.

Software's origin story starts in the 1960s, when American Airlines and IBM built a system called Sabre — moving flight reservations from paper filing cabinets into a database. That same logic then spread to every industry: HR filing cabinets became PeopleSoft and Workday, legal filing cabinets became LexisNexis, accounting filing cabinets became QuickBooks and NetSuite.

But Alex's point is this: how much did software actually improve productivity? Not that much. A company of the same size in 1950 and in 2000 had roughly the same number of HR staff. Before, someone guarded the filing cabinet so it wouldn't get rifled through. Now, there's an IT department and a CISO making sure the system doesn't get hacked. The storage medium changed. The workload didn't.

The real variable is AI. Once AI enters the picture, software is no longer just a filing cabinet — it can operate the filing cabinet. Not just show you which invoices are overdue, but call the patient directly to collect. Not just store the claims records, but submit the claims itself, track their status, and appeal the denials.

The work sitting on top of the information is orders of magnitude larger than the information itself. That's the market.

Betty Quit Two Weeks Ago. That's Your Competition.

When Lassie entered the dental market, they didn't run into a dominant software incumbent. Alex put it bluntly:

"Who's the big incumbent in dental software? We both know the answer — but it's not like Workday. Workday is a tech company; they can just ship features. In dental, the incumbent is Betty, and she quit two weeks ago. That's your competition."

For an AI software company, the most dangerous position is entering a market where a mature software player already exists — they can copy your features fast, and their users aren't going anywhere. But in industries without deep software penetration, the only competition is human labor. And human labor is walking out the door.

There are 160,000 dental practices in America, each spending an average of $200,000 a year on administrative costs. But many can't find anyone to do the work at all — so the doctor stays up until midnight doing the books themselves.

He Didn't Retire Because He Got Old. He Retired Because Betty Left.

Alex mentioned that after the Lassie launch video went out, he got a call from his father.

His father said: I talked to Ronald Sloop.

Ronald Sloop was the dentist Alex's family knew when they first moved to Florida, now somewhere in his seventies or eighties. He didn't retire because he lost his passion. He retired because after his key billing person quit, he simply couldn't find a replacement — so he sold the practice to a younger partner and walked away entirely.

He watched the Lassie launch video and sent a message through Alex's father: If this had existed back then, he wouldn't have retired.

This isn't a story about AI replacing human jobs. Lassie's co-founders kept coming back to the same point throughout the conversation: in many cases, there's no human to replace. The person just isn't there.

From Manual to 98% Automation

How did Lassie actually get to 98% automation?

Frederick's answer was direct: in the beginning, we were the humans in the loop. We took over all the work ourselves, then automated away each problem we personally ran into.

They built the context layer first — integrating with practice management systems, insurance portals, bank accounts, and patient records. Then the tool layer — API integrations that could read and write across every system. Early models weren't capable enough for complex reasoning, but the most basic automation doesn't require it. As models improved, they swapped in a stronger intelligence layer, and the product naturally got smarter.

Only after automation crossed a high enough threshold did they hand the remaining edge cases back to practice staff — and kept learning from that feedback.

Their bar was 95% or above. Not 100%, but they wouldn't ship below it. Steijn's framing: if a handful of claims still need a human touch, going from processing 200 a week down to just a few already saves the practice ten to twenty hours. Full automation is the goal, but it's not a prerequisite for shipping.

70% of Dental Practices Still Run on Paper Checks

One detail from the conversation that deserves more attention: 70% of payments at American dental practices today are still settled by paper check.

Insurance companies mail out stacks of envelopes. That scene Steijn lived through — sitting on a bar stool, opening each envelope, depositing cash and checks at the bank, reconciling every line of the itemized statement — that's still the daily reality for a huge number of practices.

Change is coming through federal regulation. The government has mandated that the industry digitize payments within a set timeline; insurers must offer electronic direct payment options. This is a regulatory inflection point, and what Lassie is doing is helping practices make that transition — collecting tax IDs and relevant information, and running the enrollment process with each major insurer on the practice's behalf.

Digitization isn't a one-click switch. Every insurer has a different system and a different process, and once you're enrolled, account permissions become a problem of their own: you generally don't want a front desk employee to have visibility into the practice's full bank statements. Lassie has productized that workflow too.

The Model Is Large, but It Doesn't Know How to Do Billing

When asked about AI's technical limits, Frederick described something they kept running into:

They assumed that because large models had been trained on so much data, they would "naturally" understand insurance billing. They didn't.

The specific workflows simply weren't there. How to submit a particular type of claim to a specific insurer. Which X-rays to attach. How to write the narrative description. Some of this knowledge doesn't exist online at all — it lives in the heads of experienced practice managers and gets passed down verbally.

Lassie is now systematically collecting standard operating procedure documents and incorporating them one by one. They have one structural advantage: they've accumulated a large volume of historical practice data and can reverse-engineer workflows from actual records.

Frederick said the technical direction he's most excited about is smaller, faster-learning models — ones that can quickly absorb domain-specific knowledge without requiring massive training datasets. The heavy product lift right now is figuring out how to capture staff preferences and working habits, and feed them into the agent in a format it can actually use to keep getting smarter.

Finding Dr. Sloop

Selling agents to enterprises means a few dinners, a contract, and seven-figure ARR. But Lassie's customers aren't on LinkedIn. They're not in Clay's database. At 7 PM, the practice still hasn't closed because an emergency patient just walked in. Back home, the dentist is ordering supplies himself because Betty left and nobody picked up those tasks.

Steijn's description: what we're doing now is mapping every dentist in America — where they are, who owns the practice, what systems they use, whether there are hiring signals like a job posted on Indeed. Then reaching them in language they actually understand.

This is a completely different GTM playbook. AI is wildly overhyped in Silicon Valley and deeply underestimated in Iowa — that's Steijn's line. Getting agents in front of practice owners in small towns in Iowa and Kentucky is, in Lassie's view, just as hard as building the product itself.

Every Small Practice Can Run Itself

Lassie's endgame isn't dental. It's every small business.

Steijn breaks it into three steps: first, capture the $200K average annual administrative cost across 160,000 American dental practices — that's a clearly visible billion-dollar market. Second, pick another practice type with a similar structure — highly fragmented, can't find staff, no entrenched software incumbent. Third, recognize that all small businesses share the same underlying logic: there's a system to read and write, there are patients or customers to communicate with, and there are books to reconcile.

The endpoint he describes: every small business owner's agent talking to consumers' personal agents, talking to insurers' agents. Before digitization, the filing cabinet was a physical object. Now, it can run itself.

Dr. Sloop probably won't come out of retirement. But the wall he hit — couldn't find anyone, had no choice but to walk away — may simply not exist for the next generation.