A low-code canvas for building complete AI applications — automations, RAG systems and LLM agents — by dragging nodes and wiring them together. The node-and-edge paradigm popularised by automation builders, applied purpose-built to AI.
Recorded from a live deployment — not a mockup.
Wire triggers, connectors and logic into a fixed automation — watch an inbox, classify each message, route it by category across Google Workspace or Microsoft 365.
Pair a model with tools, memory and data. Drop an Agent node, wire tool nodes to it, and back it with retrieval from ChromaDB or LanceDB.
Label image datasets with bounding-box, polygon and freehand tools, pre-split into train, test and valid — feeding straight into Experiments.
The studios running today include an Email Classifier, a Document Compliance checker, a Drive Upload Notifier, Calendar Meeting AI, and invoice parsers that extract data from Gmail and Outlook attachments into Iceberg.
9 document loaders for RAG ingestion, 29 built-in tools spanning search, code execution and data access, vector stores, OCR, media processors and custom code blocks.
The same credentialed connector you configure once appears here as a tool or trigger — Google Workspace, Microsoft 365, Asana and any registered MCP server.
Every studio ships with a chat tester so you can converse with an agent or fire a workflow manually before wiring it into a schedule.
Save a studio and AI Factory's Job Manager can target it — cron or plain-English schedules, with retry intervals and failure counts.