Get your AI assistant up and running in just 4 simple steps!
Set up your keys

Create or import an assistant

Drop in the Assistant Maestro element

Trigger “Maestro – Run Assistant”

Why you’ll love it
Instant feedback via custom events
As soon as the assistant finishes its reply, the assistant_finished event fires. Hook that into your UI to automatically display the answer, play a notification sound, or advance your workflow—no polling logic required on your end.

Backend workflows, no extra code
Whenever your assistant calls a function whose name matches one of your Bubble backend workflows, the plugin invokes it securely under the hood (using your bubble_api_key). You don’t have to write any glue code: upload a file, trigger a lookup, send email—just define the workflow in Bubble and the assistant will call it.

Rich file support & parsing
Drag and drop images, PDFs, CSVs, JSON or any file up to 512 MB. Vision tools automatically extract text or charts. Code-Interpreter attachments let you run arbitrary data transformations. The assistant returns parsed table data, charts, or image URLs—ready for display in your app.

Full conversation control
Override the model (openai_model) or inject a system prompt (system_message) on the fly. You get per-run metrics (prompt_tokens, completion_tokens, total_tokens) for cost tracking. And if you need to stop mid-generation, the Cancel Run action halts an in-flight request immediately.

In just a few clicks you’ll have a powerful, fully managed AI assistant driving your Bubble application—complete with end-to-end tool integration, file handling, and real-time events.