Frequently Asked Questions
Getting started
Do I need to install anything? No. The web app runs in any modern browser. There are also native apps for Windows, macOS and Linux and for Android, all sharing one account.
Is it free? In two ways. The Community Edition is free and open source (AGPL-3.0) — self-host it with Docker Compose, no quotas. On the hosted cloud, the Researcher tier is free and needs no credit card. See Editions and Pricing.
What's the difference between the Community Edition and the hosted product? The Community Edition is the core: ingestion, vector RAG chat with citations, collaborative notes, read-aloud, bring-your-own-key models. The hosted product adds deep research, the knowledge graph, AI Scientist, voice, connectors and MCP integrations, team chat and huddles, and the native apps. Full breakdown in Editions.
Documents
Which file formats can I upload?
| Kind | Extensions |
|---|---|
| Documents | .pdf, .epub, .docx, .rtf, .txt, .md |
| Spreadsheets | .xlsx, .xls, .csv, .tsv |
| Audio (transcribed) | .mp3, .wav, .m4a, .ogg, .flac, .aac, .opus, and more |
| Video (transcribed) | .mp4, .mov, .mkv, .avi, .webm, and more |
Files can be up to 200 MB each. .doc (the pre-2007 Word format), PowerPoint and HTML are not supported — convert them to .docx or PDF first. You can also import papers by DOI, arXiv or PubMed id without uploading a file.
Why is my document taking so long? Processing time tracks the work involved, not just the file size. A text-layer PDF is parsed in seconds. A scanned PDF has to go through OCR page by page, which can take minutes for a book. Audio and video have to be transcribed first. Progress is live in the UI — if it has genuinely stalled, see Troubleshooting.
Are my documents used to train AI models? No. Your documents are used to answer your questions. When you use a cloud model provider, the relevant passages are sent to that provider to generate the answer; when you use a local model, nothing leaves your machine.
Can I get my data back out? Yes — download your documents, export citations, and export notes at any time. Nothing is locked in.
Answers and citations
Where do the answers come from? From your own documents. Scrapalot retrieves the passages that actually bear on your question and asks the model to answer from those, then cites them. Click a citation to jump to the exact passage.
The answer says something that isn't in my documents. Why? Two common causes. Either retrieval did not surface the right passage — try a more specific question, or narrow the search to one collection — or the question was answered from the model's own knowledge because nothing in your corpus was relevant. Check the citations: an answer with no citation to your documents is not grounded in them.
Can it search the web too? Yes. Turn on Simple Search in the chat options to add live web results on any plan. On Pro and above, Deep Research runs a multi-phase, cited investigation across your documents, the web and academic sources.
What languages are supported? Documents and questions in many languages work, and you can ask a question in one language about a document in another. The interface is available in English, Spanish, French, German, Italian and Croatian (Settings → General).
Models
Which AI models can I use? Cloud providers (OpenAI, Anthropic, Google Gemini, DeepSeek, Groq, OpenRouter, Z.ai, or any OpenAI-compatible endpoint) with your own key, or local models through Ollama, vLLM and LM Studio. The desktop app can also download and run models on your own computer (Settings → Local AI). See Model Management.
Can I use a local model with the hosted cloud? Only if our servers can reach it. http://localhost:11434 means localhost as seen by Scrapalot, which on the cloud is our machine, not yours. Either self-host, or expose your model endpoint over a tunnel such as Tailscale.
Do I need my own API key? No. Hosted plans include the built-in Scrapalot AI model, within the plan's monthly token limit. If you prefer your own keys, Pro and Team have a BYOK price at 50% off — see Pricing. Guest accounts on the desktop app and the Community Edition are bring-your-own-key.
Accounts and teams
Can I share documents with colleagues? Yes, through shared workspaces on Pro and above. Everyone in a workspace sees its collections, the notes editor is real-time collaborative, and the workspace has a team chat. The Team plan also includes 3 seats, so two teammates get the plan at no extra cost. See Collaboration.
Can I use Scrapalot from Claude Code or another AI agent? Yes, on Pro and above. Create an API key under Settings → Integrations → Connect your agent and copy the ready-made configuration for Claude Code, Claude Desktop, Gemini CLI or the OpenAI SDK. See Integrations.
What happens to my data if I cancel? Your account moves to the free Researcher plan, whose limits then apply. Export anything you want to keep before your plan lapses. See Pricing for each tier's limits.
Still stuck?
- Troubleshooting — concrete symptoms and fixes
- GitHub Discussions — ask the community
- Discord — real-time chat
- GitHub Issues — report a bug
- support@mail.scrapalot.app — account and billing