Upload files to an agent's knowledge
Add documents so the agent can answer from them.
Where to find it
Sources › Files
Open this screen
Steps
- Open Sources › Files.
- If you have several AI agents, pick the agent in the list at the top right; sources belong to one agent.
- Drag files onto Drag & drop files here, or click to select, or click it and choose files. Several files can be added at once.
- Wait while it says Adding 1 of n…; each file then shows Added: n characters, or why it was refused.
- Check the file in the list: Ready, Empty or Failed, its size and date.
Settings on this screen
| Setting | What it does | Default |
|---|---|---|
| Upload files | PDF, Word (.docx), .txt, .md, .csv or .json. | — |
| Size | PDF and Word up to 10 MB; other files up to 2 MB. | — |
Troubleshooting
"That file is too large. Files can be up to 2 MB."
Split the file, or save it as PDF or Word (up to 10 MB).
"This file type cannot be read. Upload a PDF, Word (.docx), .txt, .md, .csv or .json file, or paste the text."
Convert the file (for example .doc to .docx), or paste its text as a text snippet.
Tips
- A file longer than 200,000 characters is cut; the row shows the part that was kept.
Good to know
- Members can read sources; only workspace managers can add, change or delete them.
Related how-tos
- Files Upload documents your AI agent answers from: PDF, Word (.docx), .txt, .md, .csv or .json.
- Preview and edit what the agent learned Read the text the agent learned from a source and fix or trim it.
- Knowledge sources: fixing common problems Fix the messages the Sources pages can show when adding files, pages and other sources.