Loading...
Written by Simas Zurauskas. Cognitive science behind the recall queue, engineering behind the live-streaming lessons, and the product principles that shape what Strive does — and what it deliberately doesn’t.
You finish a course feeling smart, then three weeks later most of it is gone. Here's what the research actually says about why that happens, and the design choice we built Strive around.
Grounding a generated lesson in a user's own documents is five stages, not one — parse, screen, assess, retrieve, attribute. Here is how each works in Strive, and the specific things it can't do.
You already have the syllabus, the slides and the readings. Turning that pile into something you can actually study from is the hard part. Here is what to give Strive, what makes a source set work, and what happens when the material runs thin.
When you upload documents to an AI product, the question underneath every other question is what happens to them. Here is the full answer for Strive — where files go, who processes them, how long anything is kept, what deletion actually deletes, and the places where the honest answer is a hedge.
Every AI learning tool now promises a personalised course. Most of the differences that matter are invisible on the pricing page. Here are the seven axes that actually separate them, and where we sit on each — including where we come off worse.
AI explanations are unusually fluent, and fluency is exactly the cue we misread as knowing something. Here is the failure mode, the evidence behind it, and a protocol for using AI to prepare for a real exam on a real deadline.
Most people don't believe retrieval practice beats re-reading until they see the numbers. Here are the three studies that nail it down, and what to do about it on a Tuesday afternoon.
A 5-minute routine for keeping what you just learned, without setting up Anki, building 500 cards, or learning a scheduling algorithm. Just a calendar and a notebook.
Most AI products show a spinner while the model writes. Strive doesn't. Here's the engineering behind streaming structured lesson blocks to the browser as they generate.
A 19th-century psychologist sat alone with a metronome and discovered the most replicated finding in cognitive psychology. Most of online learning still acts like he didn't.
We pulled the onboarding goals from 1,820 v1 signups and ran the cluster analysis. Here's what people actually want from an AI course product, and what we changed because of it.
Most "AI tutor" demos are RAG over course material. That pattern is a great fit for Q&A and a poor fit for generating coherent curricula. Here's the trade-off.