The problem: a warehouse with thousands of items where stock and location checks were done manually. Slow, data scattered across systems, errors from manual lookup.
What I built:
A system where a staff member asks a Telegram bot in plain language, by text or voice message, and gets the stock answer immediately.
Voice messages are transcribed with Whisper, then an AI layer turns the free-form request into a structured query, n8n hits the InvenTree API and returns the answer in chat. Everything runs in Docker, access is closed off through a Cloudflare Tunnel.
Models run locally on Ollama, so warehouse data never leaves the client's infrastructure.
Results:
— Manual work reduced by roughly 70%
— 10+ hours saved per week
— Around 10,000 inventory items covered
— Instant answers instead of manual lookup across systems
Stack: n8n, Docker, PostgreSQL, InvenTree API, Telegram API, Ollama, Whisper, Cloudflare Tunnel.
#n8n #automation #AI #Telegram #inventory
What I built:
A system where a staff member asks a Telegram bot in plain language, by text or voice message, and gets the stock answer immediately.
Voice messages are transcribed with Whisper, then an AI layer turns the free-form request into a structured query, n8n hits the InvenTree API and returns the answer in chat. Everything runs in Docker, access is closed off through a Cloudflare Tunnel.
Models run locally on Ollama, so warehouse data never leaves the client's infrastructure.
Results:
— Manual work reduced by roughly 70%
— 10+ hours saved per week
— Around 10,000 inventory items covered
— Instant answers instead of manual lookup across systems
Stack: n8n, Docker, PostgreSQL, InvenTree API, Telegram API, Ollama, Whisper, Cloudflare Tunnel.
#n8n #automation #AI #Telegram #inventory