Garages
A mechanic sends a voice note with a plate number. Particles answers with the exact brake pads and their stock status.
Frontier AI.
For your parts counter.
Particles turns WhatsApp messages — text, voice notes and photos — into identified parts, live availability and orders in your ERP.
WhatsApp intake: text, voice and images
License plate to VIN in seconds
Hebrew and English, auto-detected
Priority, SAP B1 and Hashavshevet connected
AI part lookup with photos from the web
Your AWS keys. Your data. Your server.
Built for every counter that sells parts.
Importers & wholesalers
Hundreds of daily WhatsApp requests become structured orders in Priority, without a single retyped catalog number.
Retail parts stores
A customer sends a photo of a broken side mirror. Particles identifies the vehicle, the part and the price — before the customer reaches the store.
Do it all with Particles.
WhatsApp Assistant
Every incoming message is transcribed, translated and parsed. Plate numbers, part descriptions and catalog numbers are extracted automatically.
Hi, I need front brake pads for 12-345-67
Found it: Mazda 3 2019, VIN JMZBP... Front brake pad set, OEM BHY1-33-28Z — in stock, ₪240. Should I reserve it?
Vehicle Intelligence
Plate to VIN to full spec card via Israeli data providers — CarCar, CarData, InfoCar — with a free government-data fallback.
- Plate
- 12-345-67
- VIN
- JMZBP...
- Model
- Mazda 3
- Year
- 2019
Catalog & ERP
Search the catalog, check availability and place orders directly in Priority, SAP Business One or Hashavshevet.
PrioritySAP Business Oneחשבשבת Hashavshevet
AI Playground
Bring your own AWS keys, pick any Bedrock model per task, and test prompts before going live. Every token is counted per user.
$ bedrock.converse(model, prompt) ▌
Your keys. Your data.
Bring your own AWS
All AI runs on your AWS account with Bedrock models you choose. No third-party AI middleman.
Encrypted credentials
Every API key you enter is encrypted at rest with AES-256-GCM. Nothing leaves your server.
Works with your ERP
Self-hosted with Docker Compose next to your existing systems. Your inventory never syncs to anyone's cloud.
Frequently asked questions
What is Particles?
Particles is a self-hosted application for spare parts shops. It receives customer requests on WhatsApp (text, voice messages and photos), extracts the vehicle license plate and the requested part with AI, resolves the vehicle VIN through Israeli vehicle data providers, finds the part in your ERP catalog or on the web, and replies to the customer with availability and price.
Which ERP systems does Particles integrate with?
Particles integrates with Priority (OData REST API), SAP Business One (Service Layer) and Hashavshevet. It can search the catalog, check part availability and place orders in all three. A built-in mock ERP lets you evaluate the full flow without connecting a live system.
Which languages does Particles support?
The interface is available in Hebrew and English. Incoming WhatsApp messages are language-detected automatically, and replies are written in the customer's language. New languages can be added with a single translation file.
How does the AI work and who pays for it?
Particles runs on your own AWS account using Amazon Bedrock. You choose which foundation model handles each task — extraction, image understanding, chat and web scraping. Voice notes are transcribed with Amazon Transcribe, which supports Hebrew. Token usage is metered per user inside the app.
How is it deployed?
With Docker Compose on your own server: one container for the app and one for PostgreSQL. A reverse proxy such as nginx or Caddy provides HTTPS, which the WhatsApp webhook requires.