Multi-agent system of AI agents for business (n8n)
Problem. In small businesses, one owner handles everything at once: leads, invoices, documents, content. Routine consumes the day, and something inevitably gets lost.
What was done. Designed a system of interacting AI agents on n8n, where each manages its own block — sales, finance, documents, content. Data flows between agents through a single database (Supabase), agents pass tasks to each other and report to the owner via Telegram. The person connects only at key points (price, contract, payment).
Result. Instead of "the owner does everything manually" — a conveyor that handles the routine itself, allowing the owner to make decisions rather than press buttons.
Stack: n8n · Supabase · LLM (GPT / Gemini / Claude) · Telegram · crypto payments · webhooks.
What was done. Designed a system of interacting AI agents on n8n, where each manages its own block — sales, finance, documents, content. Data flows between agents through a single database (Supabase), agents pass tasks to each other and report to the owner via Telegram. The person connects only at key points (price, contract, payment).
Result. Instead of "the owner does everything manually" — a conveyor that handles the routine itself, allowing the owner to make decisions rather than press buttons.
Stack: n8n · Supabase · LLM (GPT / Gemini / Claude) · Telegram · crypto payments · webhooks.