Hey PH! I'm Alessandro, and I built SARA because I was tired of seeing businesses pay thousands for chatbots that can only answer "What are your opening hours?" SARA is different — it's an AI agent with function calling. When a customer asks "Table for 4 tonight?", SARA doesn't say "Please call us." It checks availability, books the table, flags allergens for the kitchen, and sends a WhatsApp confirmation. All autonomously. The fun part: it runs on free LLM APIs. Groq → Cerebras → SambaNova → Mistral, with automatic failover. If one gets rate-limited, the next picks up. Total LLM cost: EUR 0/month. I built it as part of SCALA, an AI operating system for SMBs. SARA is the first component I'm open-sourcing (AGPL-3.0). There are 20 industry verticals already wired — restaurants, clinics, salons, real estate, gyms, automotive, and more. Each one has its own knowledge base, tools, and prompt engineering. You add your business data, not your architecture. Try it: ``` git clone https://github.com/Alessandro114... cd sara && cp .env.example .env docker compose up -d ``` Would love your feedback. PRs for new verticals are especially welcome!

SARA

Open-source WhatsApp AI agent - 20 industry-specific brains

Open-source WhatsApp AI agent - 20 industry-specific brains