無料
One question. One citation. See RecursiQ™ work.
₱0永久
- ローリング週間使用予算
- 1 プロジェクト
- 出典引用
- 100 MBナレッジライブラリ
- ドキュメントアップロード
- 50以上のAIモデル
Build your knowledge library. Ask anything. Get cited answers from your documents, the web, and 200M+ scholarly papers.
無料プランあり · クレジットカード不要 · 60秒でRecursiQ™を体験
プロフェッショナルなナレッジライブラリを構築
One question. One citation. See RecursiQ™ work.
For serious knowledge workers starting their library
無料プランの全機能に加えて:
Your AI research team. Unlimited depth, always cited.
Researcherプランの全機能に加えて:
You get vs. standalone cost:
Four things professional buyers need to know before sharing their documents.
Documents stored on Cloudflare R2 with AES-256 encryption. All traffic over TLS 1.3.
We use OpenRouter and direct API providers. None of them train on API traffic. Your documents stay yours.
Documents are private by default. Sharing is explicit and revocable. Only you can grant access.
Delete a document and it's removed from storage and all embeddings. No lingering copies.
A chat tool forgets your documents when the session ends. Tatsulok persists your documents across sessions, organizes them into collections by client or project, and gets smarter with every document you add. Your library is a compounding asset — the more you add, the more powerful every future query becomes.
Some AI tools can search web + documents + scholarly papers in one query. Tatsulok's actual difference: (1) your library is persistent and organized into collections — not a per-session file upload that vanishes; (2) RecursiQ™ cites the exact document, page, and paragraph, not just 'from your file'; (3) your library gets smarter over time as we ship new AI analysis features — it's a compounding asset, not a service.
Yes. Documents are encrypted at rest (AES-256) and in transit (TLS 1.3). Stored on Cloudflare R2. We never use your documents to train AI models — we use OpenRouter and direct API providers, none of which train on API traffic. See our Trust & Security page for full details.
If Tatsulok doesn't save you measurable time in your first month, email us and we'll refund your first month's payment. No forms, no conditions. We offer this because we believe the product works — if it doesn't work for you, you shouldn't pay for it.
The most capable frontier models: Claude 4.6, GPT-5, Gemini 3, DeepSeek V3, and more — curated for quality over quantity. You pick the model per conversation based on what you need — frontier models for deep analysis, faster models for quick lookups.
Team plans include shared collections with role-based access. One person can upload and organize; the whole team can query. Shared libraries are great for law firms, consulting teams, or research groups where everyone needs access to the same documents.
PDFs (including scanned with OCR), Word documents, Excel spreadsheets, PowerPoint presentations, plain text, images (JPEG, PNG — content extracted), and more. Video and audio transcription is also supported for recorded interviews, depositions, or meeting recordings.
NotebookLM is session-based — your library doesn't persist. Perplexity is primarily a web search tool with some document capability. Tatsulok is purpose-built for persistent, organized knowledge libraries: you build it over time, it grows more powerful with every document, and every answer cites the exact source in your library.