MindZone
Your browser's digital nootropic. Enhance your cognition with metas.A neuro-acoustic instrument that generates structured, multi-phase brainwave-entrainment audio. Each session layers binaural carriers, amplitude-modulated entrainment, and ambient synthesis into one continuous stereo render. It now adapts in real time to biometrics from a wearable, delivers sessions as audio, synchronised audio-visual, or a visual breath pacer, and can render either pure entrainment or a tempo-locked musical mix.You're about to test an early version of MindZone. You tell it what you need, it builds a session of sound, light and breathing for it, and your two lines of feedback after each session decide what we fix next. This page explains everything before you start.
What follows: the product, the engine behind it, and what we’re raising.What follows: the pipeline, the science it stands on, and the parameters.What follows: the safety notice, how it all works, and one press to start.
For partners & investors
Tell us what you need to get done. We build a session for it.
Deep work at a desk, studying for an exam, a training block, a long stretch of analysis. You type that in plain language. MindZone builds a session around it: length, intensity, and whether you get sound, light, breath, or a mix.
Built around the job you typed in
Most focus apps hand you a few moods and play a similar track for everyone. MindZone starts from what you asked for. Two people with different jobs get different sessions.
What you are browsing now is the working alpha prototype: the session engine, the four channels, and real renders. It is here to be tried and to collect feedback, not to sell. The engine already generates high-quality entrainment sessions from the task you name, whether that is a training block or an analytics shift. The raise pays for research, engine work, and testing first, and then the first version of the real product.
One system. Four ways to use it.
Same personalization underneath. Pick how you want to experience the session.
MindWave
Audio alone for focus, sleep, or winding down.
MindField
Sound and light together. Good when you want a stronger focus session.
MindSight
Light and breath lead the way. Soft reset without headphones if you want.
MindSync
Watch path that can retune the session as you go. Still in development (+$4 later).
How a session works
Personal to the ask
Length, intensity, and channel follow what you typed. You can see how the session was shaped.
Fresh audio every time
Each session is generated for that request. We aim for about 70% gross margin after render.
Our pricing
limited (free) / Core $8.99 ($69 yearly) / Core + watch $12.99 ($99 yearly). We expect most paying users to land around $96 a year.
Who it is for
People who need a clean block of focus: desk work, exams, training, analysis. A performance tool for everyday work.
Why the sessions feel structured
We borrow careful ideas from published work on attention, sound, and breathing. We keep claims modest. MindZone is a wellness and performance tool for everyday work.
- Steady sound patterns help attention settle into a rhythm instead of jumping around.
- Sessions change over time so the start, middle, and end match the job you described.
- Breath cues can calm the body when you need a softer reset.
- Light, when you want it, can sit alongside the audio for a stronger focus block.
After December 2026 we plan a university study with a research peer ($8K of the raise) to test how a session's characteristics affect motivation and follow-through. Detail lives in the Tech view.
Thirty seconds from a real session
A short clip rendered by the same system you can open next. Try the full product when you are ready.
Prototype today. A different commercial app next.
mindzone.cloud is a live research alpha. Testers (about 15 people) run real sessions and leave short notes. There are no paying customers yet, by design.
This UI, onboarding, and packaging are not the consumer product we will ship. Right now the work is research, engine quality, and testing with real people, not going live and marketing. The raise funds that work, plus design, API and hosting so sessions stay live, and runway. The generative engine stays.
Plan after the raise: keep improving the engine, run the university study with a research peer, and widen testing. Paid launch and customer-cost learning come after that work, not before it. The gate before a larger round is unchanged: about 100 payers, 40%+ still paying a month later, and CAC known within about +/-30%.
Near-term path:
- Commercial app live. Stripe (or equivalent), limited (free) / Core / Core+watch, rebuilt consumer UX.
- YouTube CAC known. Scale ads only when payback makes sense (target under about $40).
- University study after December. Better limited (free) and Core defaults. Product research, not a clinical trial.
- Seed on numbers. If we hit ~100 payers with retention and known CAC, look at $750K-$1.5M. Hardware only if the software earns it.
The numbers we plan with
Planning math for the first paid year. These are planning assumptions, not trailing results.
- Prices: limited (free) / Core $8.99 month ($69 year) / Core + watch $12.99 month ($99 year).
- Planning ARPU: about $96 per paying user per year (mix of monthly and annual).
- Target YouTube CAC: under about $40. Pause or cut ads if payback stays bad.
- Gross margin: about 70% after render. Heuristic on limited (free) protects margin if the AI path is expensive.
- Proof before seed: ~100 payers, 40%+ still paying ~1 month later, CAC known within about +/-30%.
- Lean ops burn (no ads): about $1.9-$3.1K / month (hosting, API, tools, operator essentials amortized).
English-language focus tools are a large market. Our near job is smaller and clearer: turn this prototype into a paid product people finish a work block with, at a known cost to acquire, with enough runway to survive a bad CAC month.
The ask: $120,000 and where it goes
Raising $120,000 in total, weighted toward research, engine work, and testing rather than marketing. A cheque of any size helps at this stage. Here is where every dollar goes:
- AI-orchestrated product development: $38,000 (32%). Ship the paid commercial product with agents: billing, shell, onboarding, entitlements, paid-path polish. Includes tooling / light build help.
- Design / UX: $9,000 (8%). First-run, trust UI, and visual system for the commercial app.
- API + hosting: $18,000 (15%). New LLM provider (TBD) credits plus servers, backups, staging for ~12–18 months.
- YouTube creative + ads: $15,000 (12%). Capped learning budget. First $2–5K before scaling. Pause if payback is bad.
- Research / university pilot: $8,000 (7%). After December. Stronger exploratory study. Better defaults. Not medical claims.
- Operator essentials: $12,000 (10%). Sole-employee tools, workspace, basic living buffer so execution does not stall.
- Legal + runway reserve: $20,000 (16%). Counsel plus contingency. Lean ops without ads is about $1.9-$3.1K/month; unused ad budget becomes runway too.
If only $90K closes, we keep the product build, design, API, hosting, and essentials; cut ads to about $8K, defer the university study to the seed round, and hold the reserve at about $16K. If this fits how you invest, create an account below or email us.
Investor account
One press creates an investor seat on this browser (investor1, investor2, and so on). Separate from tester seats. No password to type. Then you can open the product and try it yourself.
Alpha testing
Hear it before anyone else. Help us tune it.
We're looking for people who'll actually use MindZone for a week and tell us the truth about it. In exchange, your access never expires.
Read this first
Safety notice: photosensitivity
MindZone AI is an experimental cognitive optimization prototype. It is not a medical device, and its sessions are not clinical treatments. By participating in this free alpha, you acknowledge that the visual engine utilizes flickering light washes and agree to use it at your own risk.
If you, or anyone who can see your screen, have epilepsy or a history of seizures, do not take part. Gentle light washes are on by default; stronger flicker stays locked behind a separate in-app photosensitivity screening.
What you're testing
MindZone makes custom audio (and light, and breathing) sessions to help you focus, relax, or sleep. There are no pre-made tracks. Every session is built for you, from your own words.
The map
One press below and you land in Meta, the lobby. From there, four doors:
How testing works
The deal, in plain words:
- You get premium for life. Every channel, every future feature, free forever on your account.
- You start right away. One press of the button below creates your seat, with no password or email to type.
- After each session, tell us two lines. What you felt, what worked, what didn't. The next session unlocks once you do.
- We recognize this browser profile with a necessary cookie. It keeps one-click access stable even when your network changes. Another browser or cleared cookies creates a separate tester account. IP is kept with security, session, feedback and agreement records and is never used to decide account ownership. The full list is in the Privacy Policy.
What to write in your feedback
After each session we ask for a couple of lines. Honest beats polite. The most useful notes tell us:
- How you felt before and after. "Was stressed, felt calmer by the end" is perfect.
- What you were doing while it played. Studying, working, falling asleep.
- What the sound was like. Too loud, too dull, an annoying pulse, a nice ending. Anything you noticed.
- Anything odd. If it crashed, stalled, or felt wrong, say so. That's the most valuable note of all.
Start testing
There's nothing to fill in. One press creates your tester seat and takes you straight in. This browser profile is recognized for one-click return visits. If you run a team, class or organisation and would like a separate testing group, write to aioraownr@gmail.com.
That one press is also your agreement: you've read the safety notice above, you're 13 or older (under 18 with a parent or guardian's OK), and MindZone may set necessary login and browser-profile cookies and store your sessions, feedback, IP and browser information while the alpha runs. Do not copy, share, forge or deliberately rotate the browser cookie: it is your account-continuity credential. Privacy & Terms.
Abstract
MindZone turns a plain-language description of a cognitive goal into a structured, multi-phase brainwave entrainment session. NVIDIA Nemotron 3 Ultra (via NVIDIA NIM) reads the request and writes a strategy specification: a sequence of phases, each set by its target brainwave band, entrainment frequency, carrier offset, timbre, and breath cadence. A deterministic heuristic planner remains as the offline fallback. A signal-processing pipeline then renders that specification into one continuous stereo file, using additive synthesis, binaural carriers, sinusoidal amplitude modulation, convolution reverb, and a noise bed tuned for stochastic resonance.
The design draws on published work: the 40 Hz GENUS protocol (Iaccarino et al., 2016, Nature), the auditory steady-state response (Picton et al., 2003), and the theta-gamma cross-frequency coupling described by Lisman & Jensen (2013). A session follows a phase arc, moving deliberately between bands instead of holding one fixed tone, so the stimulus tracks the cognitive task rather than working against it.
This page covers the architecture, the science it rests on, how to run a session, and the full parameter reference. The live instrument is open to alpha testers at mindzone.cloud/meta. You can play a thirty-second rendered sample below without signing in (§ 06).
Architecture
Theoretical basis
Auditory steady-state response (ASSR)
Amplitude-modulating a carrier at frequency f produces a phase-locked cortical response at the same frequency, measurable on an EEG. It is how an audible session can entrain a target band.
EVIDENCE: STRONG
40 Hz GENUS protocol
Combined audio-visual stimulation at 40 Hz drives gamma across cortex and hippocampus and improves cognitive markers in animal models. The basis for MindZone's learning-oriented gamma phases.
EVIDENCE: STRONG (animal)
Binaural beats & the frequency-following response
Two carriers offset by Δf presented dichotically produce a perceived beat at Δf. The effect requires headphones and is weaker than monaural AM, so MindZone keeps binaural carriers ≥20 dB below the foundation pad.
EVIDENCE: MODERATE
Theta-gamma cross-frequency coupling
Working memory appears organised by gamma cycles nested within a theta rhythm. Sessions therefore couple bands across a phase arc rather than holding a single frequency, mirroring this nesting.
EVIDENCE: STRONG
Alpha entrainment & attentional gating
Alpha-band activity (8–12 Hz) indexes inhibitory attentional control. Relaxed-focus and creative phases target this band to lower distractibility without forcing high arousal.
EVIDENCE: MODERATE
Stochastic resonance
A calibrated noise floor can raise the detectability of a sub-threshold periodic signal. The pink/brown noise bed is set to support the entrainment envelope, not to mask it.
EVIDENCE: MODERATE
Slow-wave sleep & acoustic enhancement
Sound delivered in phase with slow oscillations during deep sleep can enhance slow-wave activity and memory retention. Informs the delta-targeted descent of sleep sessions.
EVIDENCE: MODERATE
Sleep spindles & consolidation
Delta and spindle activity during NREM sleep support memory consolidation. Sleep phases prioritise an unbroken descent through theta into delta over any audible novelty.
EVIDENCE: STRONG
Isochronic amplitude modulation
Pulsing a tone on and off gives a stronger entrainment envelope than binaural beats, and it works on speakers. MindZone uses a sinusoidal AM shape, never a square one, capped at 0.10 depth so the pulse never turns alarm-like.
EVIDENCE: MODERATE
Melanopsin & non-visual light response
Intrinsically photosensitive retinal ganglion cells respond to light colour and intensity, shaping alertness and circadian state. The visual engine maps colour temperature per band on this basis.
EVIDENCE: MODERATE
Empirical aesthetics of acoustic stimuli
Perceived pleasantness of a stimulus modulates engagement and tolerance over a long session. Timbre, reverb, and humanised timing are tuned so a session is bearable for its full duration.
EVIDENCE: AESTHETIC
Cyclic-sighing breath pacing
Extended-exhale breathing lowers physiological arousal faster than passive attention. MindSight's breath pacer drives an inhale/exhale ratio matched to the target state.
EVIDENCE: EXPLORATORY
Reading-friendly typography (Lexend)
The interface is set in Lexend, a typeface designed and measured to raise reading proficiency: widened default spacing improves reading speed and lowers effort for readers with dyslexia or attention and focus difficulties. It carries into the consumer product. This build is a research prototype; the full interface is rebuilt for consumers after more research, team growth, and engine work.
EVIDENCE: APPLIED
Operating manual
Parameter reference
| Parameter | Range | Default | Unit |
|---|---|---|---|
| session.duration_min | 1-120 | 45 | min |
| phase.brainwave | δ θ α β γ | n/a | band |
| phase.entrainment_hz | 0.5-100 | band-bound | Hz |
| phase.carrier_left_hz | 80-300 | 200 | Hz |
| phase.carrier_right_hz | left + ent_hz | derived | Hz |
| phase.filter_cutoff | 1200-3500 | 2200 | Hz |
| phase.am_depth | 0.05-0.10 | 0.08 | ratio |
| phase.transition_sec | 0.5-45 | 30 | sec |
| phase.intensity_scale | 0.5-1.4 | 1.0 | scalar |
| visual.color_temp_k | 2200-5500 | 3500 | K |
| visual.brightness | 0.06-0.80 | derived | unit |
| visual.dimension | 1.30-1.50 | 1.38 | D |
| visual.strobe | off | gated | n/a |
| breath.inhale_sec | 2-10 | 4 | sec |
| breath.exhale_sec | 3-12 | 6 | sec |
Live instrument
Find the instrument at:
Running it requires a login. For evaluation access during the preview, contact me; my credentials are in § 08. The thirty-second sample below plays without a login, and it is real engine output rather than a mock-up.
Changelog
Updated
- Changelog filled in from v0.9.1 to v0.13; every release now lists what was updated and what changed
Changed
Nothing changed in how MindZone works.
Updated
- MindField and MindSight: the feedback card's Send button, the Stop button and voice input now work. They were never connected, which blocked the next session
- Errors now say what actually went wrong, in all seven languages, instead of “Unable to generate a session right now”
- AI planner: one 110-second time budget, an overloaded model is skipped for 3 minutes, and a parsing bug that discarded valid plans is fixed
- Each page loads only the reader's language: the landing page went from 110 KB to 39 KB compressed
- Server: daily database backups, capped log files, and an automatic restart if every render worker hangs
Changed
- A second tester now waits in line, with their place shown, instead of being turned away
- Pressing Generate while a session is already running reconnects to it
- Fonts load from mindzone.cloud; pages no longer contact Google
- With the system's Reduce motion setting on, MindField and MindSight visuals move more slowly and the flicker stays off
Updated
- Animation cost scales to the device, so the site runs more smoothly on weaker laptops
- Scripts and styles are cached by content, so repeat visits download almost nothing
- New hero line and Invest copy; the Meta page heading was removed
Changed
- On slower machines, background blur and the pointer ring are switched off
Updated
- Headphones / Speakers choice in MindWave and MindField
- Phase changes hand the harmony over instead of overlapping two keys and two beats
- Split phases show as one row in the plan; the result card is fully translated
- MindField visuals run on the audio clock; the spectrum recovers after a tab switch; plan panels match across channels
Changed
- On Speakers the beat is delivered monaurally, in both channels, because binaural beats need a separate channel for each ear
- Skipping through a MindField session now moves the visuals and breath pacer with the music
Updated
- A hung render can no longer starve the render pool; dependency versions are pinned
- The scrubber no longer jumps back to the start while seeking
- Landing page: audio-reactive hero, band colour ladder and theater mode
Changed
- Musical sessions can run up to 60 minutes (was 30)
- Feedback on the last session is required before the next one, with no grace period
Updated
- One-press investor seats
- Security hardening: join limits, request size limits, job visibility, OAuth error handling, weak secrets rejected
- Long musical sessions no longer run out of memory mid-render
- Invest and Testing views rewritten as scroll stories
Changed
- The Free tier is now called Limited
- Each account has a daily cap on AI planning; past it, sessions use the built-in planner
Updated
- Stop generation button and wait-time estimates
- Finished sessions and job status survive a deploy or restart; fixed empty results and silent audio after completion
- Voice input in the prompt field, photos in feedback, and multi-part workouts timed correctly
- Every interface string translated into all seven languages; phone layout fixes
Changed
- A render interrupted by a server restart is marked failed instead of resuming on its own, so a new session can start
- The strategy preview under the MindWave prompt was removed
- Strategy generation moved to NVIDIA Nemotron 3 Ultra (NIM), with Nemotron 3 Super as fallback
- Deterministic intent anchoring + semantic band guards on AI strategies
- Offline heuristic path retained when NIM is unavailable
- One-press tester onboarding: a seat per browser profile, no passwords or forms
- Entry chooser on first visit; consent records for cookies & testers
- Feedback loop: the next session unlocks after a two-line note
- Admin sign-in moved onto /admin; the login page is gone
- Terms of Service & Privacy pages; 13+ age gate
- Phase-arc cross-fade extended to 30 s default
- Per-phase parallel render enabled by default (≈1.3× faster on vibrations)
- Stereo correlation target clamped 0.5 – 0.85
- Spatial-drift auto-pan added to master chain
- Visual engine: PSE gate dual-layer enforcement
- Strobe waveform changed: square → cos²
- Breath pacer morphs shape per brainwave band
- Musical mode: beat + bass + arp layer added
- Tempo-locked to band tonal centre
- Research model: visual + breath overlays per phase
- Chromotherapy hue mapping per brainwave band
- Deterministic duration parsing (hours / minutes / word forms)
- Heuristic keyword map for seven cognitive states
- Schroeder reverb + additive harmonic synthesis
- Filter-cutoff floor raised to 1200 Hz
- First end-to-end render: prompt → strategy → WAV
Authorship & contact
Designed and built by Yan Petrov.
Independent researcher and developer, working across neuroscience, digital signal processing, and accessible cognitive support. MindZone is a one-person project; the research behind it is cited inline in § 03.
For technical discussion, collaboration inquiries, or evaluation access: