Mavera DocsWorking with files: what you can feed it
Everything you can upload, what happens to each type, and the workflows they unlock.
Turn a file into working evidence
Upload, route, inspect extraction, analyze, and save a grounded result.
Use the sidebar to open the workspace file library.
Good afternoon
Choose a surface from the sidebar to begin.
- 1Open FilesUse the sidebar to open the workspace file library.
- 2Upload a sample fileDrop in the included sample interview transcript.
- 3Choose what it becomesRoute the file to Context, an audience ingest, or analysis.
- 4Inspect extractionOpen extracted pages and verify what was read.
- 5Analyze itAsk the sample question against the uploaded evidence.
- 6Save the resultSave the grounded output to the project.
Mavera eats documents. Here's the full menu of what to feed it and what you get back.
What you can upload, and where
| File type | Best place to drop it | What it becomes |
|---|---|---|
| Pitch deck / one-pager (PDF, PPT) | Context → Upload & extract | Business facts Mave uses everywhere |
| Customer interviews, call transcripts | Ingest | Grounded audiences that quote like real customers |
| Survey exports, CRM lists (CSV/Excel) | Ingest or Analyze | Audiences (Ingest) or charts + takeaways (Analyze) |
| Reviews scraped from anywhere | Ingest | Voice-of-customer audiences |
| Brand guide, logo | Context → Brand kit | Extracted palette + on-brand image generation |
| Contracts, RFPs, reports | Files or the chat paperclip | Read, summarize, compare, extract, edit |
| Screenshots (ads, competitor pages) | Chat paperclip or Context extract | Described and folded into the conversation |
| Any doc mid-conversation | The paperclip in Mave's composer | Instantly readable in that conversation |
The three file superpowers
1. Files → knowledge (Context)
One good deck ≈ ten solid business facts, extracted and categorized in about a minute. You approve each fact before it saves. This is the highest-leverage 5 minutes on the platform.
2. Files → people (Ingest)
The current Ingest workflow is for completed research studies. It creates a source-quoted dossier for review and analysis. It does not yet turn a folder of interviews, tickets, NPS verbatims, or reviews into a reusable grounded audience.
Use only completed-study material you are authorized to upload, and review the current storage and deletion limits first.
3. Files → analysis (Analyze / chat)
Drop a spreadsheet and ask questions in plain language:
What's driving the drop in week 6? Chart revenue by segment. Which customers look like churn risks?
Charts and takeaways come back computed from the actual data — and it will say "the data doesn't contain that" rather than guess.
Size and format limits
Up to 6 files per extraction batch, 8 MB each on Context. The Files page takes documents up to 500 MB. Excel, Word, PowerPoint, PDF, CSV, images, and plain text all work. Scanned PDFs work too (they're read visually).
Prompt starters for file work
Attach a competitor's pricing page screenshot: "Compare this against our pricing. Where are we vulnerable?"
Attach two versions of a proposal: "What changed between these, and which is stronger for a risk-averse buyer?"
Attach a CSV of survey results: "Find the three most surprising things in here. Then build an audience of the unhappy segment."