One agent, five tools: an AI-run back office

Here's the pattern our users keep converging on, whether they sell real estate, waterproofing rolls or gym memberships: one conversational agent, wired to the boards that run the business.

The setup

Take a sales agent for a real-estate company. Its flow looks like this:

Chat/WhatsApp Input → LLM Agent ⇄ CRM (leads)
                              ⇄ Kanban (deal pipeline)
                              ⇄ Inventory (listings)
                              ⇄ Analytics (metrics)
                              → Chat Output

The five connected Tools aren't decorations — they're function-calling tools the model invokes on its own. When a customer says "I want to visit the two-bedroom on Saturday", the agent registers the lead in the CRM, moves the deal card to Visit scheduled, checks the listing's availability, and confirms — in one reply.

Why this beats a chatbot

A chatbot answers questions. This operates: every conversation leaves structured records humans can supervise from the board UIs. And when a case needs a person, the Chat tool's human takeover pauses the AI and hands the thread over.

Build yours in an afternoon

1. Create the agent with a clear objective.
2. Create the Tools you need in Apps → Tools.
3. Wire them into the LLM node and enable function calling.
4. Publish the chat — web widget or WhatsApp.

Every step is documented in the Help Center, and Genius can scaffold the whole thing from one message.