File Processor Plugin (Agentic Orchestrator)
The Agentic Orchestrator is in Controlled Availability and comes to your Enterprise Assistant in Q1 2027. Capabilities, limits, and interface labels described here may change before general availability. To request access, contact your Moveworks account team.
The File Processor plugin is how the Agentic Orchestrator answers questions about files. A user attaches a file and asks about it, and the assistant reads the file and answers. There’s no “Ask questions about file(s)” button, no mode to enter, and nothing to exit, so every other plugin stays available in the same conversation.
It does the same job Brief Me does on the Agentic Reasoning Engine.
How the Assistant Decides to Use It
The model calls the plugin itself when a question is about a file, so there’s nothing for users to learn: they attach a file and ask. It doesn’t call it for a general question just because a file is in the conversation.
- When it’s called. The question is about a current attachment, names a file or an earlier upload, or continues a discussion about a specific file.
- One file per call. Each call reads one file. If a question spans three files, the assistant calls the plugin three times, once per relevant file.
- What it can read. Files attached to the current message, files uploaded earlier in the same conversation, and artifacts the assistant created earlier.
- When it’s unsure. If the assistant can’t tell which file the user means, it asks instead of guessing.
- What the user sees. A step that reads “Querying the selected file” with the question, then the answer.
- File content is data. The assistant treats text inside a file as document content. It does not follow instructions written in the file.
For exact math over a spreadsheet or CSV, the assistant can use Code Execution instead. The model picks between the two based on the question.
Limits
Upload limits are shared with Code Execution. See Code Execution.