Used by 100+ students in 10+ countries and counting
A two minute film. It opens with the gap I watched appear in my own classroom, follows one class from the first question to the review page, and shows the evidence behind each design decision.
Learnica captures the class and keeps the page moving, then asks you to predict, choose, or recall at a few useful moments. You get the ease of Otter or Granola while more of the thinking remains yours.
Start class
Choose the microphone for a live room, the browser tab for a lecture playing elsewhere such as Coursera, or both at once. A YouTube link opens inside Learnica the same way. Every path lands in the same live workspace.
Before class
Learnica asks for a quick prediction before class begins. The question is optional, but answering it gives the coming explanation something to connect to.
In class
Live captions, translation, and a complete transcript catch the words you miss. The page stays as flexible as GoodNotes or Notability without making you chase every sentence.
While notes generate
Every AI note taker needs a moment to turn a recording into notes. Learnica uses that same pause to ask what you can recall before the finished page opens.
After class
The concept mirror compares your recall with what the class covered and points to what needs another look. Otter, Granola, and Notion AI show what was said, while Learnica also shows what reached your own words.
In class
One tap marks a passage as important or as something you did not follow. The yellow and purple markers stay on the heading, so review starts where you actually struggled.
Any time
The assistant works from the class record and from your own notes. It opens with a question or the smallest useful hint, and it gives the full explanation once you have made an attempt.
The gate
A line of your own, a stroke, or a captured slide opens the assistant and the notes. Learnica also asks for a minute of class and a few words of your own, so an empty keystroke is not mistaken for thinking.
The same task can help one learner and overwhelm another. Learnica treats support as a hypothesis, then looks past the finished page to unaided recall, transfer, and calibration.
It depends on what the learner already knows, what the task asks, and what must still be possible when the tool is gone.
Prediction, selection, retrieval, and self-explanation ask the learner to construct or recover an idea, so Learnica keeps those moves in the flow.
Racing to transcribe, recovering lost audio, and decoding a second language can consume capacity without strengthening the target idea, so Learnica carries that load.
These are design judgments, not permanent labels. The assistance dilemma is deciding when support removes waste and when it removes the thinking that produces learning. Prior knowledge, timing, scaffolding, fading, and expertise reversal can change that answer. Learnica tests each change against delayed unaided recall, transfer, metacognitive calibration, and the learner's experience instead of treating a cleaner artifact as proof of learning.
Each one has a plain interaction at the surface and a testable mechanism underneath.
A quick guess gives the coming explanation somewhere to land, even when the guess is wrong.
Headings and prompts remove the fear of a blank page while leaving the meaning for the student to build.
Bringing an idea back strengthens it in a way that reading the same idea again does not.
A clear mirror helps students separate an idea they understood from one they only remember seeing.
A finished note can look better while learning gets worse. The product therefore separates performance with help from what remains when the help is gone.
Sample inference item · one week laterA clinic wants answers grounded in this week's guidelines. Why would retrieval serve it better than fine-tuning?
Two studies make the same tradeoff visible. A polished result can improve while the learner becomes less able to work without the tool.
The page can look complete in all four. The difference is who had to keep up and whether the learning process is visible afterward.
A fluent response is the easy part. The system has to survive the room, retrieve the right evidence, release only the help the learning moment allows, and show where it came from. Learnica separates capture, evidence, policy, generation, and evaluation so one polished answer cannot hide a broken source or a missing learning step.
The live product captures the class, shapes notes from class materials, and compares recall. The next build adds a tool-using tutor that releases help in stages.
Learnica asks the learner to commit to one idea before the lecture. The answer stays locked, while capture and translation remain available.
More context can bury the passages that matter, so the model may follow the wrong evidence.
Today, help waits for a thought of your own. The planned engagement check will look for meaningful language rather than count characters.
Complete captions and translation follow the room while the note stays short enough to belong to the student.
A capable model can still answer badly when the context is weak.
Retrieval first finds relevant passages, then adds them to the prompt as evidence.
More context is not automatically better. Retrieval and generation need separate evaluation.
RAG lets a model look things up before answering, but the retrieved context still has to be relevant.
Learnica keeps the learner's shorthand visible beside additions grounded in class sources. The next provenance layer will make the source of each addition directly inspectable.
RAG lets a model look things up.Retrieval finds passages before generation.The passages become evidence in the prompt. [1][2]Retrieval and generation need separate evaluation. [3]The tutor retrieves and verifies course evidence first, then chooses the smallest useful response.
Matched the question to the transcript at 18:42 and the RAG diagram on slide 12.
Retrieved the component and end-to-end evaluation framework from slide 28.
Course evidence was sufficient, so web search did not run.
def retrieve(query, course, k=4):
hits = lexical(query, course, k=20)
if low_confidence(hits):
dense = semantic(query, course, k=20)
hits = rrf(hits, dense)
return rerank(query, hits)[:k]One trace follows a live class through an audio failure. The other follows a request for help through the planned tutoring policy.
A student may switch devices, lose a headset, or pass through a short network interruption while the lecturer keeps speaking.
+00:00The headset stream drops mid-sentence.+00:02Unsent audio chunks are held in IndexedDB on the device.+00:09Capture resumes on the new source, inside the same session.+00:11Captions and translation rejoin at the sentence boundary.The policy chooses a learning move before any answer is drafted, so orchestration does not quietly become answer generation.
Learnica keeps streaming when an audio source changes or the network drops briefly, then returns captions as complete sentences with translation moving beside the lecture.
The current gate asks for a thought of the student's own before AI help becomes available. The next version will check for meaningful engagement rather than count characters.
The deck, transcript, and concept list define what generated notes may cover. The student's own words remain distinct from the generated record, and the source material stays available for review.
Capture, gates, and release remain deterministic. Planning begins only when the learner's attempt, the available evidence, and the help policy leave more than one valid next move.
Each student receives a server-held condition for each lecture. If that state is missing, the intervention stays closed rather than assigning a condition on the client.
An exact course term should be found quickly. Semantic retrieval, reranking, graph traversal, or the open web enter only when the question needs them, and remembered claims stay visible to the learner.
Before Stanford, I taught more than 1,000 undergraduates across seven computer science courses. As generative AI coding assistants became part of everyday coursework, I sometimes saw elegant take-home code, and then, on an in-person final, a simple question we had practiced many times was answered wrongly, or came back blank. There was no need to assume cheating, because the work could be genuinely theirs while too much of the thinking still happened somewhere else.
At Stanford, I began treating that classroom question as both a learning-science problem and a product problem. I brought years of building full-stack AI systems at Apple, TikTok, Alibaba, AMD, and early-stage startups to a tool that now serves about 100 students. Learnica keeps the complete record of a class, follows the room with captions and translation, and creates a few well-timed opportunities to predict, choose, recall, and explain. The aim is to make a difficult class easier to enter, wherever a learner begins, without letting a polished artifact stand in for understanding.
Education becomes personal when a teacher notices the hesitation behind a polished page, remembers an earlier struggle, and stays long enough to help the idea land. Learnica can carry the record, keep the evidence close, and make more room for those moments. What it cannot do is care whether the idea finally lands, because that responsibility and that relationship must remain human.
Built to be tested, not just believed.
Zikun "Zayden" Zhu