Walk into any busy dental office at lunchtime and you'll find the same quiet problem. The phone is ringing, someone is checking in, a claim needs filing, and somewhere in the practice's software there is a list of several hundred people who are overdue for a cleaning. Everyone means to call them. Almost nobody does. That list - the patients who drifted off, the treatment plans that were diagnosed but never scheduled - is where the money goes to die. 10x Dental was built to work that list, and to do it with an AI that never gets pulled away to the front counter.
The company, part of Y Combinator's Winter 2025 batch, describes itself plainly as "AI sales teams for dental practices." In practice it connects to a practice's existing patient database, identifies three kinds of missed revenue - overdue hygiene recalls, lapsed patients, and unscheduled treatment - and then reaches out by phone and text to bring those people back into the chair. When a patient says yes, the system writes the confirmed appointment straight back into the office's scheduling software. No spreadsheet, no manual re-entry.
The problemThe ledger nobody reads
Dentistry has an unusual kind of leak. Practices rarely lose patients in a dramatic way. Someone moves, someone's insurance changes, or someone simply doesn't rebook after their last cleaning. Six months later they are a line in a database that nobody opens. Multiply that by a few hundred names and the number gets serious fast - practices routinely leave five figures a month on the table, not because the demand isn't there, but because the follow-up work never gets finished.
The reason is not laziness. It's arithmetic. A front-desk team that is already answering live calls and running check-ins cannot also dial four hundred overdue patients a day, handle each person's "let me look at my schedule," and log the result. So the reactivation work gets pushed to a someday that never comes. 10x's bet is that this specific task - high-volume, repetitive, easy to script, and endlessly deferred - is exactly the kind of thing an AI should be doing.
It helps that the timing is unkind to dental front offices. Practices across the country report that hiring and keeping front-desk staff has become one of their hardest operational problems, which means the phones get worked even less than they used to. A dropped recall doesn't announce itself the way a broken chair or a no-show does; it just shows up months later as a schedule with holes in it. That invisibility is precisely what makes the leak so easy to ignore and so expensive to leave alone.
The numbers16 versus 3
The figure 10x leans on most is conversion. According to the company, a human front desk converts roughly 3% of overdue recall calls into booked appointments. Its AI converts about 16% - and books more than five times as many overdue patients as human staff over the same period. The gap isn't really about the AI being a smoother talker. It's that the AI actually makes all the calls, every day, without tiring of the list.
Figure: More than 5x the booking rate on the same overdue list, per 10x's reported data.
Those percentages turn into dollars quickly. In case studies the company points to, practices booked somewhere between roughly 38 and 125 patients in a month and recovered between $11,000 and $33,000 in additional production in their first month of use. One practice in Miami is generating more than $30,000 a month in incremental revenue. A practice in Los Angeles brought back 94 lapsed patients in a single month.
The productThree jobs, one AI coworker
10x sells its work in three modules, and the split tells you how the company thinks about a practice's day. There's Hygiene Recall, the daily outreach to overdue patients with live booking. There's Unscheduled Treatment Follow-Up, which chases down the treatment plans that were diagnosed but never put on the calendar. And there's Missed Call Answering, which calls a patient back when the front desk can't pick up, handles the objection, and books the slot - closing the leak where a new patient calls at lunch, gets voicemail, and dials the next office on Google instead.
All of it runs two-way. Patients can talk to it or text with it, and the whole thing is HIPAA-compliant with a business associate agreement for practices that need one. A dashboard tracks the daily tally - recalls worked, production recovered, conversion rates - so the office can see what the AI actually did rather than take it on faith.
The moatBoring plumbing that wins
The least glamorous part of 10x is probably its most important. The product integrates directly with the practice management systems dentists already run - Dentrix, Open Dental, Eaglesoft, Curve, Dentrix Ascend, CareStack, Denticon, iDentalSoft, Archy, Oryx and others - and writes confirmed appointments back into them automatically. In healthcare software, the tool that already holds the data tends to win, and the newcomer that can speak its language without asking staff to export a CSV gets adopted. That integration list is the real barrier to entry, not the voice model.
The foundersFrom grading papers to dental chairs
10x's two founders, Max Bohun and Aman Garg, did not start in dentistry. Their previous company was GradeWiz, an AI teaching assistant that graded and supported coursework and was used in Ivy League classes - the kind of tool where, as they put it, one TA could do the work of ten. After roughly two years, they wound GradeWiz down and, inside the same YC cycle, relaunched around a completely different market. The through-line is the thesis, not the industry: take a repetitive, human-bottlenecked task and let AI do it at volume. A small fossil of that history survives in the company's email, which still runs on the gradewiz.tech domain.
The team is deliberately small - a two-person founding group out of San Francisco, hiring its first founding engineer and go-to-market hire as it grows. That constraint shows up as focus. Rather than spreading across ten features, 10x has pointed its energy at one measurable outcome and gone looking for practices that will pay for it. The early traction suggests the approach is landing: the company is live in the largest and second-largest privately owned dental groups in Chicago, a major Sacramento practice, and dozens more nationwide, and says it is doubling month over month.
The modelPriced on the money it makes
The pricing is the tell. At $970 a month for the flagship recall module, 10x isn't charging on what the software costs to run - it's charging a fraction of the production it recovers. When a practice pulls $11,000 to $33,000 out of a dusty ledger in the first month, a four-figure subscription reads less like an expense and more like a rev-share the customer happily rounds up on. It's a B2B SaaS sold per module, with discounts for multi-location groups and the dental service organizations that run them - a natural fit for a market of roughly 200,000 dentists in the U.S., most of whom are wrestling with the same front-desk labor crunch.
The field around it is filling in. Passive patient-communication platforms have existed for years, and a fresh wave of AI-native dental startups - handling everything from insurance verification to billing to full front-office automation - arrived at roughly the same moment 10x did. That crowding cuts both ways. It validates that dentists will pay for software that touches their revenue, and it means 10x has to keep proving that "we book the patient" beats "we message the patient." Its answer so far has been to stay narrow and let the recovered-dollars figure do the arguing.
There's a broader shift underneath all of this. Voice AI has, in a remarkably short span, gone from a demo-day novelty to a line item a dentist describes as their highest-ROI integration. 10x didn't try to automate the expert in the room. It automated the bottleneck around the expert - the phones nobody had time to work. Whether that turns into a durable business or an early move in a crowded field, the pattern it's riding is hard to miss.