Start Here, In This Order
About thirty minutes of reading and one account, and you will know whether this is worth your time.
The About section has sixteen pages and no obvious front door, which is a problem this page exists to fix. Below is the order we would read them in, what to do once you have an account, and what the first week actually looks like. If you came here to build on the platform rather than use it, skip to the last section.
Read these three pages first
Three pages, in this order, will give you enough to judge the rest.
1. Overview
What V.E.T.S. is and why it is built the way it is. It explains the loop the whole platform runs on: the AI writes a first draft, an expert corrects it, and the correction becomes what the next person reads. If you only read one page, read this one.
Start with what the platform is →
2. Core Concepts
Five or six words that the other fifteen pages use constantly and never stop to define: TeamDoc, the tree, a minion, a chunk. Ten minutes here saves you re-reading later.
Learn the words the rest of the section uses →
3. Use Cases
Written accounts from clinics, breeding operations, farriers and livestock managers. Find the one closest to your own work and read that first. Abstract descriptions of knowledge management are much less useful than one concrete account of a Tuesday.
Find the practice closest to yours →
Create your account
Signup is at the link below. You need an email address; there is nothing to install.
Create your account →
Everything in this About section is readable without an account, which is deliberate. The pages that describe the architecture, the AI layer and the security model are public because a buyer should be able to evaluate them before handing over an email address. What an account gets you is the working application: your own records, your own tenant, and the assistants that operate on them.
What the first week looks like
Day one: put something real in
Create a handful of genuine records rather than test data. The platform is more useful, and considerably easier to judge, when it has something true to work with.
Day two: notice which assistant answers
Fourteen assistants, called minions, are configured across 78 page contexts. You do not choose between them. The screen you are on determines who responds, so Florence answers on patient history and Penny answers on invoices without you selecting anything.
Meet the assistants and see how routing works →
Day three: correct something
This is the part that matters, and it is the part new users skip. When an assistant produces something wrong, edit it. That correction is the contribution — not a bug report about the AI, but the corrected record itself.
One seam worth knowing about, because it will otherwise look like a fault. Your correction is saved immediately and is visible to you at once. It is not immediately retrievable by the AI. A scheduled job, ajob_AI_GCPVectors, carries new and changed content into the vector index every hour, on the hour. Until its next pass, the assistant may still answer from what you just fixed.
All week: use it where the work happens
Nearly every clinical record type has a _Mobile counterpart in the procedure layer. Of the 33 farrier procedures, 27 are astp_Farrier_Mobile_*. That proportion is not an accident of naming; the system was built on the assumption that the record gets created in a barn aisle on a phone, not that evening at a desk.
The same assumption is behind the microphone button in the minion panel: tapping it starts Drive Mode, which listens, sends what you said and speaks the reply back, so you can ask a question with both hands busy. How Drive Mode works, and what it will not do →
If you came here to build on it
The reading order above is written for someone who will use V.E.T.S. to manage animals. If you are evaluating it as an engineer, that order wastes your first hour.
Start instead with the developer entry point, which covers the naming conventions, the split between sacred and application objects, and how to find your way around 2,834 stored procedures without reading them.
Get productive in the codebase instead →
Where to go next
If you would rather see the whole thing work end to end before committing thirty minutes to it, there is a page that follows one case from the first record to the answer that comes back.
Walk through a real case end to end →
Ready now? Create your account.