Veterinary ·Confidential, US veterinary supply platform
The clinic's next order is already in the cart, built from what it actually bought before.
US veterinary clinics where the buyer is the practice administrator or the vet, re-keying the same repeat orders by hand across scattered vendor lists.
Where this fits
Best for vertical marketplaces whose buyers are practitioners rather than professional purchasers, ordering the same things on a cycle.
The next order is predicted from the clinic's own history. 1 year in production, 15+ vendors connected.
- Elasticsearch catalog search
- Event-sourced order history
- One-click GPO connection
What we did
- Read a clinic's order history and predicted what it needs next
- Assembled an optimised order across 15+ US veterinary vendors
- Built a GPO layer with membership terms and one-click connection
- Kept the full order change history on an event-sourced model
Result
The administrator opens a cart that is already close to right instead of rebuilding it from memory every week. It has run for a year across 15+ connected vendors, and clinics save several hours a week on purchasing.
The hard part
The prediction is the easy part. Repeat purchasing looks like forecasting and is really reconciliation: the same product carries a different SKU at every vendor and again inside each GPO contract, pack sizes differ, and which price applies depends on a membership the clinic may or may not hold. Get the mapping wrong and the clinic either receives the wrong pack size or loses the contract price, and notices on the invoice rather than in the app. That is why order history is event-sourced: when a vendor disputes a line, you have to replay exactly what was ordered and on which terms.
- Stack
- JavaSpringPythonReactNext.jsNode.jsPostgreSQLElasticsearchEventStoreKubernetes
- Focus
- Vertical AIProcurement AutomationPredictive OrderingVeterinary