Lassie

Posting insurance payments is the job Lassie automates. It pulls explanations of benefits from payer portals, matches each payment to the patient, the claim and the bank deposit, and writes the result to the ledger with the practice's adjustments and ledger notes. Lassie puts the share of posting done without a person at 98 percent and says the system needs less input as it learns a practice.

The company now pitches agents for the rest of the back office: EFT enrollment, conversion from paper checks, denial follow up, claim status calls and month end close. Its engineering listings describe AI agents built on frontier models.

Most payments post without a person; the rest come back to the office. Practices set posting rules down to how each adjustment and allocation is handled, Lassie flags receivables and other issues for review, and it asks the practice for input on the most complex cases. Posting experts back up the software by WhatsApp, phone and email.

What turns a posting into an exception, and how a payment posted to the wrong ledger is caught and reversed, are not described.

Lassie's engineering job listings say more about the technology than its product pages. They describe AI agents built on frontier models using the Pydantic AI framework in Python, evaluations as part of the engineering work, and a browser sync engine on encrypted SQLite for patient ledgers.

The product pages say only that Lassie is AI native and learns each practice's posting habits. Which models read the EOBs, how accuracy is measured and how changes are released are not published.

Engineering job listings put Google Cloud Platform and Vercel in Lassie's stack, alongside frontier models, and its agreements bind subcontractors to the business associate agreement and let third party products such as payment processors pass data into the platform. None of the model providers or subcontractors is named.

EOBs carry patient identities, procedures and payments, and every party in that chain handles them.

Customers supply most of the proof, and none of it comes with a baseline. Dr. Eric Kwon of Grace Dental reports more than 100 hours of data entry saved in a month and outstanding receivables down 75 percent. The owner of Pediatric Dentistry of San Jose says EOBs that took a full afternoon now take 10 minutes, and Dr. Nelson Kanning says his best team member no longer spends hours entering checks.

Lassie's own figures are broader: 98 percent of posting automated, payment four times faster, more than 15 hours a week saved, and onboarding in one to two weeks, or about eight for offices still on paper checks. None says how it was measured or over what period.

Practice data can be reused well beyond the service. The business associate agreement lets Lassie use and keep insights from the service, and de identified information it owns, to improve its products, including for other customers, and to run data aggregation. The services agreement adds that aggregated, anonymized datasets built from client data may be used for any lawful purpose, including platform optimization and testing. Whether any of it trains a model is not said.

At the end of a contract, protected health information is returned or destroyed where feasible, the practice has 30 days to export patient data, and Lassie may keep a copy as long as the law requires.

The business associate agreement is published and takes effect with each client services agreement, so every practice is covered from signup. It binds subcontractors that handle protected health information to substantially similar terms, requires written breach notice within 30 calendar days of discovery, and returns or destroys the information at termination where feasible.

Its permissions are wide: Lassie may use protected health information for data aggregation and to create de identified data that it then owns. The website privacy notice covers marketing and site data only and leaves patient data to the agreements.

Lassie's services agreement commits to commercially reasonable administrative, physical and technical safeguards, and the company walks prospects through its security in a demo. No outside audit or certification is named.

The footprint inside a practice is deep. Offices install a Lassie server on their own Windows machine and a remote access tool on a server or workstation, and Lassie connects to payer portals and the practice's bank account for reconciliation. On its side, the engineering stack includes a sync engine on encrypted SQLite in the browser.

Payment posting, reconciliation and payer enrollment are financial administration, and Lassie makes no medical device claim.

Where it bills, Lassie acts as the practice's agent, in the practice's name and under the practice's collection policies, and it may endorse insurance checks for deposit. The services agreement leaves the practice responsible for its fee schedule and for the completeness, accuracy and timeliness of the documentation behind each claim.

Lassie's engineering job descriptions list evaluations as part of the work on its posting models. No evaluation of its posting decisions is published.

Posting errors land on patients as wrong balances on their statements. A system that reads some payers' EOBs less accurately than others would concentrate those errors on the patients of those plans.

Every posting lands on the patient ledger with a note, receivables problems are flagged for review, and posting experts handle questions, which gives a practice a trail to find and reverse a bad entry.

The contract carries the risk the other way. The platform is provided as is, Lassie's liability is capped at the fees paid in the prior 12 months, its indemnity covers only intellectual property claims, and as billing agent it does not guarantee how much will be collected. No error rate or correction commitment for postings is published.

Dentrix, Open Dental and Eaglesoft are the named systems, and Lassie writes into each through its installed server: postings land on the patient ledger with the adjustments and ledger notes the practice's rules call for, with no exports or uploads. Payments are reconciled against the bank account as they arrive.

Other practice systems are mentioned without names, and no integration documentation or marketplace listing is published.

Delivery combines a hosted service with software installed at the practice: a Lassie server plus a remote access tool for a server or workstation. The platform links to payer portals and the practice's bank account.

The engineering stack runs on Google Cloud Platform and Vercel, according to Lassie's own job listings, but no hosting region or data location is stated for patient data. The agreement is governed by California law, and patient data can be exported for 30 days after it ends.

Billing follows the work: Lassie charges when it does work for a practice and quotes a price after reviewing the practice's volume and workflows; no rate, tier or figure appears on the site. Getting started runs through a demo.

The services agreement fills in the mechanics. Fees are set in each order, may include onboarding fees, are non refundable unless the order says otherwise, and are collected monthly on a date Lassie can change.

The typical customer is an independent dental practice on Dentrix, Open Dental or Eaglesoft. Lassie's landing pages say more than 1,000 dentists use it, its company page says more than 700 doctors' offices, and its job listings say thousands of practices. Owners of general, family and pediatric practices vouch for it by name, including Grace Dental, Infinity Dental Associates, Kanning Dental and Pediatric Dentistry of San Jose.

The pitch is widening to every doctor's office, but the product pages, onboarding and agreements are written for dental practices. Larger groups and DSOs get no separate offer.