Built on memory science

Built to
remember.

Hebbr is a memory-science study app. The techniques that feel like learning usually aren’t — we use the ones that are: encode, retrieve, teach back, space, each walked by an AI partner.

Try Hebbr free How it works

A card’s journey

01
Plan
Drop a topic, syllabus, or URL — Hebbr proposes a single deck or a multi-deck course plan.
02
Generate
Claude drafts the cards, a second pass fact-checks each one before it lands.
03
Loop
Four phases over time. One Study button — Hebbr picks the right phase per card, every session.
EncodeRetrieveTeach-backSpace
04
Exam
Practice exam mode tests what stuck. Weak cards loop back through the study phases.
05
Mastered
Cards graduate. FSRS keeps them spaced just before forgetting — so they stay learned.

The science

Built on how memory actually works.

A card’s journey

Four phases, across days — not one session.

Every card moves through the same loop over time — the phases memory science says actually wire a fact down. Each session, Hebbr picks the phase each card needs today. One button does the picking.

01Tutor

Encode

A new card surfaces. In your own words — what is this like? You produce the hook; Hebb refines it and saves it with the card.

02Tutor on miss

Retrieve

When a card comes due, recall it cold. Hebb grades, explains what you missed, and moves on.

03Protégé

Teach-back

Once you’ve retrieved a card, explain it back to Hebb. He plays the student — asks a question, points out the gap, makes it stick. The act of teaching is what locks it in.

04Silent

Space

FSRS schedules the next touch — the moment before forgetting, the card comes back. You don't pick when to review.

hebbr.app · one session

hover to pause · click a phase to jump

EncodeMake it yours first.Tutor

New card

Chlorophyll

In your own words — what does this remind you of?

Like an antenna — it tunes into specific wavelengths of light.

You

Nice — the antenna lands. It hands the signal off to the reaction center.

✓Hook saved to card

RetrievePull it cold.Tutor on miss

Card due · 1 of 12

What is chlorophyll?

The green pigment that absorbs light and drives the first stage of photosynthesis.

AgainHardGoodEasy

Close — it specifically absorbs red and blue, reflects green.

Teach-backExplain it to Hebb.Protégé

Chlorophyll soaks up red and blue light and passes that energy to the reaction centers, which kick off the light-dependent reactions.

You

Wait — so a “reaction center” isn’tjust chlorophyll? What’s actually at the center doing the reacting?

Gap surfacedreaction center mechanism

SpaceScheduled for the slip.Silent

FSRS places the next touch at the moment recall is about to slip.

Upcoming reviews

Today

+1d

+2d

+4d

Card

+7d

+14d

FSRS forecast94% retention at +4d

The evidence

Every phase, a principle behind it.

Spaced repetition

Ebbinghaus, 1885 · Bjork, 1994

Review a card the moment it starts slipping and the synapse fires just in time to restrengthen. Too early and nothing fires — wasted effort. FSRS predicts that moment per-card.

Active recall

Roediger & Karpicke, 2006

Retrieving from memory activates the same circuits that store it, and the act of retrieval itself strengthens them. Practice Review and Practice Exam force retrieval on every card.

Elaborative encoding

Craik & Lockhart, 1972

Memories wired to more context are wired to more retrieval cues. The AI Tutor explains the why around a card, not just the what — so the fact lands in a web of meaning.

Metacognitive calibration

Dunlosky & Metcalfe, 2009

Knowing what you know is its own skill. Rating yourself Forgot, Struggled, Knew it, or Nailed it after each card trains that judgment — and gives FSRS the signal it needs.

The generation effect

Slamecka & Graf, 1978

Producing an answer yourself encodes it more durably than reading one. With AI Protégé you teach the concept out loud; with AI Peer you argue it through. Both force generation — and generation is where wiring happens.

Pre-testing effect

Richland & Kornell, 2009

Guessing before you know primes the brain to encode the answer when it lands. Hebbr asks for a confidence rating before every reveal — committing to a guess, even a wrong one, makes the answer stick harder than passive exposure would.

Desirable difficulty

Bjork, 1994

Retrieval that's harder in the moment is stickier in the long run. Practice Review's free-recall mode — type or speak your answer from scratch — pushes past the shallow recognition you'd get from multiple-choice. The added effort is the point.

Dual coding

Paivio, 1971

Verbal and visual memory live on parallel channels — a fact wired through both is wired twice. When you forget a card, Hebbr generates a visual mnemonic alongside your written hook, giving the next retrieval two routes to the same answer instead of one.

Interleaving

Bjork & Bjork, 2011

Mixing topics within a session forces the brain to discriminate between patterns, building more distinct traces. Hebbr's cross-deck review stitches due cards from every deck into one session — biology, Spanish, and legal cases in the same sitting, on purpose.

The protégé effect

Roscoe & Chi, 2007

Teaching forces you to reorganize what you know — and the gaps surface as you explain. Hebbr unlocks Teach-back the moment you’ve retrieved a card once; you don’t wait for mastery, you build it by explaining.

Recall over time

stylized

Without reviewWith Hebbr

Hebbr uses FSRS — the Free Spaced Repetition Scheduler, a modern open-source successor to SM-2 — to predict the moment each card is about to slip, and schedules it there. You spend less time reviewing. You remember more.

Your AI study partner

One AI. Three roles. The loop picks.

The best way to learn something depends on where you are with it. A tutor when you’re stuck, a student when you think you’ve got it, a peer when you want to think out loud. Hebbr picks the role that fits the phase — you can override with one tap.

💬Tutor

Explains what you don't get.

“Think of FSRS as a weather forecast for your memory — let me show you why that matters here.”

Elaborative interrogation

🧑‍🏫Protégé

Learns from you — on purpose.

“Wait, so spaced repetition works because forgetting is… good? Can you say that part again?”

The Feynman technique

🗣️Peer

Thinks alongside you as an equal.

“I keep mixing up SM-2 and FSRS — want to trade: you take the first, I'll take the second?”

Socratic dialogue

From topic to deck

Three steps to a deck you can trust.

Describe what you want to learn

Drop a PDF, paste a URL or YouTube link, or just type a topic — Hebbr can read what you bring or pull the course material off the web for you. Either way, it proposes a single deck or a multi-deck plan grounded in real sources, not guesses.

Generate cards you can trust

Claude drafts the deck, a second pass fact-checks every card, and flagged cards arrive with a suggested fix you review before saving.

Hand it to the loop

Your deck drops into the four-phase study loop. Hebbr schedules each card for the moment its recall starts slipping and runs an AI partner through every session.

About the name

Named for Donald Hebb.

In 1949, Canadian psychologist Donald Hebb described the rule behind modern neuroscience: when two neurons fire together, the connection between them strengthens. That principle — Hebbian learning — is the biological reason memory sticks. Every phase of the loop is built to fire the right cells together at the right moment.

Hebb’s rule also seeded artificial intelligence. In 1982, John Hopfield built the first associative-memory neural network on Hebbian learning — work that earned a share of the 2024 Nobel Prize in Physics with Geoffrey Hinton for the foundations of modern machine learning. The same rule that wires your memories wires today’s neural networks.