Add Documents to Your Memory Dataset
Your assistant answers from its memory dataset. Crawling a website is one way to fill it; uploading documents is the other. Use uploads when the knowledge lives in files — a handbook, a price list, a policy PDF — rather than on a public page.
Upload Files
- Open the assistant's dataset, or go to Memory datasets and open the one you want.
- Choose Import → Upload files.
- Add your files (or drag them in). Remove any you picked by mistake.
- Start the upload and wait for processing to finish.
Files are processed, not just stored
Seekdown extracts and indexes the text so the assistant can search it. A newly uploaded file becomes answerable after processing completes — not the instant it finishes uploading. If new content isn't found yet, give it a moment and refresh.
Supported Formats
PDF, Word (DOCX), Excel (XLSX), PowerPoint (PPTX), plain text (TXT), rich text (RTF), Markdown (MD), and ZIP archives. If a file is rejected, check its extension and size against the upload dialog.
Beyond uploads, the same dataset can also grow through a data capture job or the content API (an Enterprise feature) — a dataset can mix all three.