DEV Community

Ali
Ali

Posted on

How we built an AI-powered veterinary clinic management SaaS (lessons learned after 50+ clinics)

Hey DEV Community πŸ‘‹

We're a small team from Istanbul, Turkey, and I wanted to share our journey building VetAgent β€” an AI-powered veterinary clinic management system that now serves 50+ clinics.

The Problem We Solved:
Veterinary clinics in Turkey (and most of the world) still use paper records or clunky legacy software from the 2000s. Vets spend 30-40% of their time on documentation instead of treating patients.

What We Built:

  • Voice-to-SOAP Notes: Vets press a button, describe the visit, and AI generates structured clinical notes (Subjective, Objective, Assessment, Plan)
  • Lab Device Integration: Direct connection to IDEXX, Fuji, and ASTM lab devices β€” results flow automatically into patient records
  • 30+ modules: Appointments, pharmacy, POS, invoicing, vaccinations, surgery planning, inventory β€” all in one platform
  • Mobile app built with Flutter for on-the-go access

Tech Stack:

  • Backend: Laravel 11 + PostgreSQL
  • Frontend: Next.js 15 + React 19
  • Mobile: Flutter 3.x
  • AI: OpenAI GPT-4 for clinical notes, custom RAG pipeline for drug interactions
  • Deployment: Docker + Hetzner

Key Lessons:

  1. Start with the workflow, not the AI. We spent months shadowing vets before writing a line of code.
  2. AI accuracy in medical contexts needs to be 99%+ or don't ship it. We use structured prompts + validation layers.
  3. Lab device integration was the hardest technical challenge β€” ASTM protocols are from the 1990s.
  4. Hardware integration (lab devices, printers, scales) is a massive moat.
  5. Turkish market first gave us fast iteration cycles before expanding globally.

Results:

  • 50+ clinics onboarded
  • ~40% reduction in documentation time for vets
  • 98% client retention

If you're building in healthtech/vettech or any regulated industry SaaS, happy to answer questions!

Website: vetagent.io | Built by: benai.ai

Top comments (0)