On a Telegram message, runs a multi-model AI agent with Postgres memory and Supabase vector search, replying across Telegram and Gmail.
On a Slack or chat message, a RAG agent retrieves from Pinecone using OpenAI and Anthropic models and replies via Gmail.
On an incoming Gmail or chat message, embeds Slack content into Pinecone and answers queries with OpenAI and Anthropic agents.
When files are added to Google Drive, embeds them into Pinecone with Gemini, letting a chat agent answer questions over the documents.