Handwrite on an infinite canvas. Get quizzed into memory. Replay your own thinking over time. Built for the way memory actually works.
iOS· Android· macOS· Windows· Linux· Web· End-to-end encrypted· GDPR-native
The bottleneck
In 2026, knowledge takes three seconds to retrieve and three weeks to forget. We have ten apps for taking notes and none for keeping what we learn.
Highlighting feels like studying. Re-reading feels like studying. Asking an LLM for a summary feels like studying. None of them are — and the cognitive science has been clear about this since the 1970s.
The bottleneck of learning is no longer access — it's encoding, retention, and transfer: three problems at once.
Fluera is built for those three.
The method
Every Fluera feature lives at a step in this loop, and every step is grounded in published research. Skip one, the chain breaks. Keep them all, the notebook becomes permanent.
Write concepts by hand during the lecture. Position them in space. Compression — forced by the slowness of the pen — is where encoding begins.
Close the book. Rebuild what you remember on a blank canvas. The red nodes where you falter are the map of what to study next (Kapur, productive failure, 2008).
The AI asks; you answer. Rate your confidence one to five before each reveal. Metacognition made explicit.
Ghost Map overlays your reasoning against the ideal. Hypercorrection makes the gaps permanent (Butterfield & Metcalfe, 2001).
Do nothing. Slow-wave sleep replays the day. The hippocampus hands traces to the neocortex. Our job is to get out of the way.
Come back the next day. The canvas fades what you almost knew. Pull it back from memory, then rewrite what slipped.
Visit a classmate's canvas. Teach. Be taught. Recall duels. Organising for others cements it for you (protégé effect).
Successive relearning at widening intervals. The gold-standard evidence-based strategy (Rawson & Dunlosky, 2011).
Zoom out to continent view. Draw arrows between chemistry and differential equations. Transfer is the real test of learning.
Fog of War mode. Navigate the mist of your own canvas. Green, red, blind spots — the mastery map makes itself.
Fluera generates a closed-book exam from your own canvas. Rate your confidence one to five, answer by hand, then Ghost Map overlays the gap. Hypercorrection makes overconfident misses permanent (Butterfield & Metcalfe, 2001).
The canvas persists as a cognitive autobiography. Return in two years and see the tangible shape of how much you've grown.
Four cognitive differentiators
After you finish a problem, Fluera overlays your work against an ideal solution. The mismatches don't shame — they shock the memory awake. The wronger you were, the more the right answer sticks.
Hypercorrection effect — Butterfield & Metcalfe, 2001.
Rate your confidence one to five before each reveal. Metacognition is half of mastery, and confidence calibration is how you build it — one prediction at a time.
Metacognitive monitoring — Flavell, 1979.
Mask any region of the canvas. Pull it back from memory before revealing. The retrieval struggle — not the seeing-again — is where memory hardens.
Desirable difficulties — Bjork, 1994.
Confidence picker one to five before each question. Closed-book. After you submit, Fluera overlays your answer against the ideal and applies hypercorrection where you were most overconfident. The bigger the surprise, the harder it sticks.
Hypercorrection — Butterfield & Metcalfe, 2001 · Productive failure — Kapur, 2008.
Ink, not type
Mueller & Oppenheimer, Princeton 2014 · van der Meer, NTNU 2020
When Mueller and Oppenheimer ran their now-famous Princeton study in 2014, they found that students who took notes by hand outperformed laptop-takers on conceptual questions — even though laptop-takers wrote significantly more words. The hand was slower. The mind was deeper.
A decade later, Audrey van der Meer's EEG work at NTNU (2020) gave us the why: handwriting recruits broader brain networks than typing, especially in regions tied to memory formation and sensorimotor integration. The pen, it turns out, is a cognitive instrument.
Fluera is built around this finding. Thirteen brush engines. Pressure, tilt, velocity. Sub-15-millisecond stroke latency on every supported device. The infinite canvas isn't an aesthetic — it's the substrate your memory has been asking for.
Scientific foundation
Every design decision traces back to a paper we can link to. Twelve of those minds are below. The full bibliography lives in a public vault.
Desirable difficulties — the counterintuitive finding that harder study produces better retention.
1994
Hypercorrection Effect — confident errors are remembered most permanently once corrected.
2001
System 1 vs System 2 — fast intuition versus slow deliberation in thinking and memory.
2011
Zone of Proximal Development — the narrow band where scaffolded practice beats solo practice.
1978
Self-efficacy — the single strongest predictor of academic success, stronger than IQ.
1977
Growth mindset — praising effort over talent changes how learners respond to difficulty.
2006
Flow — peak cognitive performance when challenge matches skill and feedback is immediate.
1990
Concept mapping — building a map beats studying a pre-built one, across disciplines.
1984
Place and grid cells — the brain's built-in GPS. Why memory palaces work.
Nobel 2014
Handwritten notes beat laptop notes on conceptual recall — even when the laptop notes were more complete.
2014
EEG evidence that handwriting recruits broader brain networks than typing.
2020
Retrieval practice — testing does not measure memory, it creates it.
2006
We cite five decades of memory research. We have not run our own randomized trials on Fluera itself — not yet. When we say "evidence-based," we mean the underlying principles are; our specific implementation is an engineering bet informed by the evidence.
We also don't promise expertise. Meta-analyses (Macnamara, Hambrick & Moreau, 2014) found that deliberate practice accounts for roughly 4% of the variance in educational performance — the rest is prior knowledge, motivation, working memory capacity, and opportunity. Fluera optimizes the practice loop. The hours, the focus, and the curiosity are still yours.
Six platforms, one ink
Native renderers on Vulkan, Metal, OpenGL, Direct3D 11 and WebGPU. Your notebook syncs end-to-end encrypted, or stays offline. Your call.
Metal
Metal
Metal
Direct3D 11
Vulkan
OpenGL
WebGPU
Status labels are honest about release readiness. Platforms marked "Alpha" or "Demo" work but need polish — join the beta if you want to help shape them.
Your mind, your data
SQLCipher encryption on every device. Your local database is unreadable without your key.
EU-based, data processing agreement available, full export and deletion on request.
No tracking pixels. No behavioural ads. We don't train our models on your content.
For Education accounts, every access is logged in an immutable trail.
Pricing
€0
The full canvas. Three notebooks. Forever free.
€5.99/mo
Unlimited canvas, 2-device sync, baseline AI. €49/year.
€11.99/mo
Exam Session, Atlas, full AI suite, unlimited sync. €99/year.
€19.99/seat
Shared canvases, team Atlas, ACL, audit log. 5+ seats.
Verified students get Plus or Pro at 50% off through our pilot program. Whoever subscribes now locks the rate for the lifetime of the subscription.
Join us
We onboard by cohort so every team — students, knowledge workers, faculty — gets a real conversation, not a queue ticket. If you're tired of apps that promise to learn for you, request access.
Request access →