Reasonote, Open-Source AI-Generated Courses, Lessons, and Podcasts
Hi HN! Luke, founder of Reasonote, an Education Tool that makes personalized AI-Generated Courses, Lessons, and Podcasts. Today we’re making it Open Source :) Try it now: https://reasonote.com/app Github: https://github.com/Reasonote/reasonote ------ The Problem We're Solving Education today suffers from two issues: 1. Traditional education uses a "one-size-fits-none" approach that ignores individual learning needs 2. Most edtech is still linear, impersonal, and doesn't adapt to student goals or interests ------ How Reasonote Works Reasonote solves these…
What it does
In the maker’s words, at launch
Hi HN! Luke, founder of Reasonote, an Education Tool that makes personalized AI-Generated Courses, Lessons, and Podcasts. Today we’re making it Open Source :) Try it now: https://reasonote.com/app Github: https://github.com/Reasonote/reasonote ------ The Problem We're Solving Education today suffers from two issues: 1. Traditional education uses a "one-size-fits-none" approach that ignores individual learning needs 2. Most edtech is still linear, impersonal, and doesn't adapt to student goals or interests ------ How Reasonote Works Reasonote solves these problems by creating personalized AI-Generated interactive courses — including Skill Trees, Lessons, Activities, Feedback, Assistance and Interactive Podcasts — to scaffold the process of understanding. There are a few key ideas that make Reasonote unique. Some of them: 1. Fully Generative, Grounded In Sources: Unlike other edtech platforms that rely on pre-written content, every aspect of Reasonote's learning experience is generated by AI, while staying grounded in your source material. The AI adapts the content and difficulty based on your goals, creating an individualized learning journey that evolves with you while ensuring accuracy. 2. Skill Tree & DAG SRS Algorithm: Unlike Anki, Duolingo, or Quizlet — we map prerequisite relationships between concepts (as a Directed Acyclic Graph) and combine this with spaced repetition. This means you'll learn concepts in-order, and when you’re ready. For instance — you’ll learn the concept of velocity before the concept of acceleration, because our system knows it's a prerequisite and you're prepared. 3. AI-Generated Activity Abstraction: We've gone beyond basic flashcards to prevent "overfitting" in learning. Our system generates eight different activity types (roleplay, sequencing, term-matching, etc.) that move past memorization to ensure deeper conceptual understanding. Developers can create new activity types by specifying three components: a generator (TS), a renderer (React), and a grader (TS). Each component has access to AI calls to make the experience as interactive as-is desired. Imagine uploading an ML paper. Reasonote would analyze it, build a skill tree of concepts (from basic stats to neural nets), create interactive lessons, and generate activities that test both understanding and application. You might find yourself in a roleplay activity explaining backpropagation to a “fellow student”, or sequencing the steps of gradient descent. As you progress, the system adapts based on your performance, ensuring you're neither bored nor overwhelmed - balancing structured mastery with the freedom to explore tangential interests. ----- Why Open Source? Education is far too important to be locked behind proprietary walls. We’ve seen the bad that came from “enshittification”, and we've seen the good that came from Wikipedia and other open-source tools. We want to build a robust future for AI-powered learning. We believe education should foster both mastery of existing knowledge and the curiosity to question it - creating learners who can stand at the frontier of knowledge and then push it forward through first-principles thinking. Finally, we believe that making the product community-driven can give us a much broader perspective, to ensure the product is useful in a wide array of domains and circumstances. ------ Roadmap We’re still early! Currently, you can upload documents (papers, syllabi, etc), and Skill Trees, Lessons, Activities, Feedback, Assistance and Interactive Podcasts are all generated for you. We want to make the product faster, more in-depth, and more intuitive. Soon, we'll support: - Learning from videos, audio, and webpages - AI-generated diagrams using Mermaid - More activity types (You can also author your own!) Thanks for checking it out! Please share any feedback in the comments :)
Does the same job
all alternatives →
- LOLearnawesome.org – Open-source learning map for humanity2020 · ▲216
Hello HN, Over the last 12 months, I have been building https://learnawesome.org This idea came from Danny Hillis' 2012 talk at OSCON, which itself was inspired from Neal Stepehenson's _The Diamond Age : A Young Lady's Illustrated Primer_ The dream is to build a tool that matches the right learning material to the right student at the right time. Wikipedia is great, but it doesn't do a good job of leveraging the rich variety of learning resources that exist on the Web. Same applies to GoodReads - which is focused only on books, whereas these days we actually learn from videos,…
- LALurnby, a tool for better learning, is now open source2022 · github.com · ▲162
I've been working on Lurnby for 2 years. It's kind of like a mix of pocket + kindle + anki. It lets you => add add epubs, pdfs, and web articles to the app => highlight and add comments => tag and organize highlights => review them with a spaced repetition system Today I made the decision to open source the project. I'm passionate about helping other people learn to learn better and hope that this will allow a lot more innovation in the tool and the space. I'm very new to open source and development in general really, but looking forward to receiving the guidance of the community.
More ai this month
the category →
I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.
AI · 17d ago · simedw.com
Astute▲585Automate your B2B brand going viral, with new media creators
AI · 18d ago · company-app.joinastute.com
Hey HN, Henry from Cactus here! We previously released Cactus Needle, a 14MB agentic LLM for tool call, device use, and structured extraction for phones, wearables, smart homes, small robots and microcontrollers. We got really great feedback here, and have now incorporated the suggestions to release Needle 2. The whole model is a single 14MB binary that runs a full session in 28MB of RAM; 45m parameters at 2bit compression. Needle hits 500 tokens/sec decode speed on a Raspberry Pi 5, sits between 400-1,500 tokens/sec on VR devices like Meta Quest 3S and Apple Vision Pro, and ranges…
AI · 26d ago · cactuscompute.com
Launched alongside, May 2025
the whole month →
- C9
Life & fun · 2025 · felixrieseberg.github.io
