
ThinkTotem
Turn boring books into engaging conversations
Hey Indie Hackers! I’m Claudio, a solo founder building ThinkTotem, an AI reading coach that transforms any document into an interactive chat so you actually remember what you read.
🤕 The pain
Passive reading is brutal: pages blur after 10 minutes, it’s hard to spot the key ideas, and there’s no feedback loop to check if anything stuck.
I kept skimming business books, zoning out, and realising weeks later that I couldn’t recall a thing.
💡 The solution
Upload a PDF, EPUB, web article, Word doc or even a YouTube link (≤ 4 MB for now).
ThinkTotem:
Maps the core concepts, trims filler and builds a “knowledge graph” of the book automatically
Runs an active-recall loop: Socratic questions, explain-it-back prompts and spaced-repetition refreshers keep you engaged and test retention
Tracks mastery so you can skip, revisit or mark chapters “done” once you’re confident
Result: most users master a 300-page business book in 1-2 hours of chat instead of days of passive reading.
🛠️ Under the hood
Ingestion ≈30 s for a 300-page PDF on Gemini Flash; cost $0.05–$0.10
Chat logic, TTS and STT run server-side on Next.js; OpenAI handles voice (~$0.40 per user-hour)
No vector DB needed—context windows are built from summaries + the most relevant chunk of the current chapter
A tiny policy model classifies each turn (summary / question / recall) so the conversation stays on track
💸 Business model
Free tier: 3 uploads, 1 h chat, 50 messages
Paid plans just scale limits—no feature gating. Standard $30/mo, Pro $75/mo, Ultra $240/mo for teams.
🔒 Privacy
Original files are deleted after processing; only derivative summaries + metadata are stored. Nothing is used for model training and you can purge everything anytime.
📈 Current status
MVP is live – you can try it now without a credit card.
~100 beta users, 0.8h average session length, 52% return within a week.
Long-tail docs (academic papers, RFCs, dense proofs) get the most love because the tool “untangles without dumbing down”.
🙏 Ask to the IH community
Does the active-recall flow feel helpful or does it get in the way?
For founders shipping courses / newsletters: would a white-label “interactive reader” mode be useful?
What would make ThinkTotem a daily habit for you?
I’m here all day—happy to share the full infra cost breakdown, walk through the ingestion pipeline or brainstorm growth experiments. Fire away! 🚀
About
We've all been there - you start reading an important book, but after 10 minutes your mind wanders and nothing sticks. ThinkTotem solves this by transforming passive reading into an active, conversational experience that

4 Comments
This is a very useful tool, Claudio. Though there are already existing platforms like Jira allowing its users to build interactive agents based on the respective organisations' database, you have defined a niche market for yourself by focusing education.
Though I see how this tool can be adding value to Notion and obsidian platforms, I would refrain from that in the early stages of your business. It is better to carve a niche market in education sector, expand into e book integrations and slowly remove the friction of upload by user. This gives you an advantage of coming to your platform if they need a book as well.
It has good potential, keep building. My best wishes.
I agree with you. Books should be the focus for now.
Thanks for the encouragement
This is super impressive, Claudio. love how ThinkTotem tackles the real problem of passive reading head-on. The active-recall flow sounds particularly useful for retaining dense material, especially for business books and technical docs.
At Tech For Ever, we’ve been following tools that blend AI and education, and ThinkTotem stands out with its clean execution and user-focused design. The white-label “interactive reader” idea could be a hit for creators offering deep-dive content, imagine newsletters or eBooks becoming live learning sessions.
Bookmarking this. Would love to see integrations with note-taking tools like Notion or Obsidian in the future. Keep building!
Thanks you a lot! Right now, any public webpage with a URL can already be imported (e.g. blogpost). But for Notion or Obsidian, it should be a public page to work.
Btw, when you use it, please send me feedback. I would be happy to improve it following your lead