🧠

My Brain

Not another app to fill in.
A second brain that remembers everything, keeps it safe in two places, and answers anything you ask.

Designed by Sandeep Karnati

0+
features, and growing
0
areas of your life
0
memories, always in sync
0%
yours — private by design

↓ scroll, or tap the chapters above

The problem

Every app organizes your tasks.
None of them understand your life.

A normal app

  • ☑️ Stores what you type
  • 😶 Forgets how the day felt
  • 📦 A filing cabinet — things go in, silence comes out
  • 🔁 Day 1000 works exactly like day 1

My Brain

  • 🎙️ You speak — it remembers, forever
  • 🔍 Ask anything — it answers from your own life
  • 🔐 Passwords, emails, people, notes — one safe place
  • 📈 Day 1000 knows 999 days more about you

The whole idea, in one line

Everything you capture goes into one brain — kept safe in two places — and the brain gives it back as answers, memories and stories — and keeps your secrets safe.

Next chapter shows it happening to a single thought. Keep going. →

Chapter 2 · how it works

Follow one thought

No diagrams to decode. Just watch one little thought travel all the way through the system. Tap any stage, or let it play.

💭“call Diksha Friday”

you say it once — then forget it. The brain doesn’t.

🎙️
1 · Capture
your voice becomes text — instantly
💭 here
🧠
2 · Understand
AI works out what you meant — quietly
💭 here
🗄️
3 · Remember — into two twin memories
saved twice, kept perfectly in sync
💭 here
📚 Your server
private memory
☁️ Cloud twin
backup brain
↩️
4 · Comes back — right when you need it
one thought, three ways home
💭 here
a task,
for Friday
📖
a line in
your story
💬
an answer
when you ask

▶︎ Playing… watch the thought move down. Tap any stage to explore it.

That’s the entire architecture. One thought in, four stages, and it lives in your brain forever — ready to come back.

and it’s not just to-dos

🔐
“the studio wi-fi is sunrise-42” → a fact, safe in your vault, ready any time
👤
“Diksha is our new architect” → a person your brain now knows
💬
“what did I decide about pricing?” → an answer, pulled from your own days

Same machinery, anything you give it — a task, a password, a person, a question, a whole meeting.

For the curious — the real stack
A React progressive web app + a NestJS (TypeScript) API in one Docker image, behind a Caddy reverse-proxy with automatic HTTPS. Operational data in SQLite (Prisma). The two “twin memories” are RAG — a Postgres + pgvector store on your own server — and SuperMemory, an independent cloud store. Every save goes to both through a retrying outbox queue, with a reconcile sweep so they can never silently drift apart. AI is called only for the moments of intelligence; it keeps nothing.

Chapter 3 · capture

Speak your mind. Literally.

No forms, no fields. Talk for a minute and a whole day assembles itself — work and family, sorted, prioritized, timed.

60 seconds of speaking → a day that runs itself

🎙️ 7:02 AM — you, speaking:

✨ your day, one second later:

📌 Send the revised quote to the architect [💼 work · high · 30m]

○ Book the dentist for Amma [🏠 personal]

○ Pick up the school project supplies [🏠 personal]

○ Review the vendor contract [💼 work · 45m]

🎙️

Voice everywhere

Hold-to-talk on every screen, with live transcription and an AI tidy-up. Morning dump, night story, a quick note — all by speaking.

📥

Capture anything

Paste text, drop a file, import from Google Drive. It’s summarized, filed into memory, and gets its own private chat.

✈️

From your pocket

Forward a link to Telegram, send a voice note, reply to a reminder — it all lands in the same brain.

Chapter 4 · how it remembers

Remembered twice. Never lost.

Everything you save is written into two independent memories at once — one a memory engine we built ourselves, running on your own server — and quietly kept in perfect sync. If one ever falls behind, it heals itself.

a memory being saved…

idle
📚 RAG · your server
☁️ SuperMemory · cloud twin

What flows into your memory — and how often

✅ Tasks · live 📥 Notes · live 💡 Ideas · live 🎙️ Meetings · live 📖 Stories · on update 📧 Emails · nightly 🔐 Vault · labels only — secrets never indexed

📚 RAG — our own memory engine (on your server)

We didn’t rent this — we built our own meaning-aware memory engine, running on hardware you control. Everything you save is searchable by what it means, not just keywords — and it never leaves your box.

☁️ The cloud twin (SuperMemory)

A second, fully independent memory — redundancy by design. A queue keeps both in lock-step so a memory can never be lost.

And you hold the controls: a Settings → Index panel shows every section, what’s indexed, when it last synced, live RAG-vs-SuperMemory counts — with a one-tap Reconcile now. Turn any section off and it’s purged from search.

For the curious — the sync engine
Every save is queued and dual-written to both stores by a retrying outbox worker (about every 5 seconds). A reconcile sweep every 15 minutes is the safety net — it re-queues anything missing from either store and retries failures, so the two can never silently drift. RAG is a Postgres + pgvector store on your own server; SuperMemory is an independent cloud store. Content is chunked and embedded; an optimizer can re-chunk older documents.

Chapter 5 · memory & search

Ask your brain, with receipts

It searches by meaning, not keywords — and every answer cites the real things it came from.

Ask, and watch it light up

A live, illustrative simulation — tap a question.

🔖 Video: “Technical SEO in 2026”
🌙 Story · the launch evening
📄 Doc: pricing research
🔖 Article: keyword clustering
💡 Idea: newsletter engine
🌙 Story · the family Sunday
🔖 Guide: local SEO checklist
📄 Doc: vendor comparison
For the curious — search internals
Your question becomes the same kind of meaning-vector and is matched across both stores; a scope filters by tags, results merge + re-rank, and the answer is grounded in the retrieved passages with sources attached. Explore (Ask · Everything · Saved), a ⌘K global search, and a per-document chat all ride the same engine. An index manager shows what’s indexed per section, when each last synced, and RAG-vs-SuperMemory counts — with one-tap reconcile.

Chapter 6 · talk to your brain

Just ask

No menus, no folders to dig through. Ask in plain words — it works out where to look, reads your memory, and answers with the receipts.

🌐 Everything 📊 My life 🔖 Bookmarks
what did I owe Diksha?
🧭 deciding: this needs a memory search → searching your brain
📖 Story · Jun 14 ✅ Task · open 🎙️ Meeting · client sync

in the real app, tap any source chip to open the exact item it came from.

🎯

Scoped or whole-brain

Ask inside one area — bookmarks, your life, a single document — or let it search absolutely everything at once.

🔗

Answers you can open

Every answer cites its sources — tap one to jump straight to the real task, story, email or doc it came from.

💬

It keeps the thread

Suggested follow-ups, saved answers, and a private chat on any single document — plus ask by voice, or ⌘K from anywhere.

Chapter 7 · the daily loop

Your day, written for you

One gentle loop, every day. Tap a step.

🌅
Morning dump
60 seconds
Your day, built
AI-crafted tasks
🌙
Evening story
your words
📖
Story of the Day
auto-written
🧭
Mentor speaks
honest, nightly

👆 Tap a step to see what happens there.

One story · two lives

It separates work from family — and watches the balance

The client call you dreaded turned into the win of the week — they signed. You closed three threads you’d been circling for days, and it shows: you ended work lighter than you started it.

Work mood
Home mood

🧭 Mentor: “Work is winning at 78 while home sits at 52 for the third week. The proposal can wait one evening — Thursday’s dinner cannot.”

📜 A longer memory

Daily stories roll up into weekly reviews, monthly chapters, and a Story of the Year — your life, written as a book.

🧭 An honest mentor

Focus areas, an on-track score with a trend line, a nightly note, and a 4 PM nudge to land the day well.

👥 It learns your people

Who matters keeps surfacing — names, context, and how your weeks revolve around them.

Chapter 8 · the lab

It doesn't just remember you —
it understands you

Every night, My Brain studies your day the way a scientist studies data — and slowly, carefully, learns who you actually are. Your own personal scientist. 🔬

📊

It reads your whole day

What you did, what you kept putting off, what you skipped — plus your story, your mood, your emails and your meetings.

🕳️

It learns from what you avoid

The richest signal about a person isn't what they do — it's what they defer and abandon. The Lab weighs your inaction the heaviest.

🔗

It connects acts to feelings

It builds a living map of what you do ↔ how you feel ↔ who's involved ↔ when — the real wiring of your days.

So it notices things like — "money & admin tasks drain you, and you keep deferring them" · "Saturdays are your deep-work sanctuary" · "you avoid conversations where you have to hold someone accountable."

Each night it shows you what it understood — and you teach it

Tap ✓ if it's right, ✗ if it's not. Your taps are the experiment. (go on, try it 👇)

Money & admin tasks drain you — and you keep deferring them.

55% · seen 6×

Saturdays are your deep-work sanctuary — your best output lands then.

62% · seen 5×

You avoid conversations where you must hold someone accountable.

55% · seen 11×

🌱 Patterns that live

Each insight carries a confidence % and its evidence. It grows stronger every time it recurs — and fades on its own rhythm when it stops (a weekly pattern won't fade on a Tuesday). Old phases quietly retire; the real you stays.

🔬 The science of you

See it all in The Lab: a living graph of you ↔ feelings ↔ people ↔ tasks, your mood trend, what energizes vs drains you, and heatmaps of your week and your avoidance.

It's honest — no flattery, just what the evidence shows. It's private — your patterns are yours, and you can pin or delete any of them. And nothing is ever "known" about you that you didn't confirm.

For the curious — how the model thinks
Each night a reasoning pass (Claude Sonnet by default — or your own free Codex / Gemini agents) reads the day's signals and proposes evidence-backed hypotheses about you. A separate lifecycle pass reinforces, decays and merges them on their own cadence, promoting the durable ones to “established.” Everything lives in both of your twin memories, and your ✓ / ✗ feedback trains the next run — so it gets truer every night.

It doesn't just hold your life — it understands you, and keeps understanding you better.

Chapter 9 · the vault

Your secrets, sealed shut

A full password manager built in — and built so that even the app can’t read it.

🔒

Zero-knowledge

Everything is encrypted in your browser with your master passphrase before it’s ever stored. The server only ever holds an unreadable blob — no one can decrypt it but you.

👆

Unlock with a touch

Face ID / fingerprint unlock on your device, a forced Recovery Key at setup, and auto-lock the moment you step away.

🙈

Never indexed

Only the labels (the site name) are searchable — the secret itself is never added to memory, never sent to any AI.

…and it’s a complete manager, not a side feature

11 secret types

logins, cards, bank, crypto, identity, API keys, docs, licenses, Wi-Fi, memberships, notes

Strong by default

password generator + weak / reused detection + a security view

Copy that vanishes

tap to copy a password — the clipboard auto-clears in 30 seconds

Extra-guarded

re-enter your passphrase before a seed phrase or private key is ever shown

Move in easily

import from 1Password, Bitwarden or CSV — encrypted in your browser

Encrypted files

attach a passport scan or doc — sealed the same way

Tidy & fast

type tabs, table or card views, collections, favorites, audit log

Encrypted backup

export the whole vault — still only readable with your key

Chapter 10 · works with your world

Plugged into your real life

It meets you where your day already happens — and treats every connection like a vault.

📧 Gmail, distilled

A nightly Daily Brief of what actually mattered, an in-app thread viewer, and important emails recalled by your brain. Connected through Google’s own secure sign-in — you stay in control, and can switch it off and purge anytime.

🔗 Noticed in your brain

A daily pass quietly finds non-obvious links across everything you’ve saved — “this idea connects to that meeting three weeks ago” — and surfaces them on your home screen.

🎙️

Meetings

Record & transcribe, then an AI summary with takeaways and decisions — searchable forever.

🤝

Commitments

“I’ll send it by Friday” is caught automatically and tracked until it’s done or dropped.

📅

Calendar & Drive

Your day’s events at a glance and one-tap import of Docs, Sheets and Slides into memory.

Tap a connection to see its role.

👆 Tap a connection.

Chapter 11 · make & know

Your workshop

A second brain isn’t only memory — it’s where you make things and keep what you learn. Capture a thought, grow a rough idea into a real plan, save what matters, and let it record the rooms you’re in.

📥

Capture anything

Paste, type, dictate or import — a link, a doc, a stray thought. One box; the brain files it.

📝

Notes

Quick notes and checklists — searchable forever, never lost in a folder.

🔖

Bookmarks

Save links with a summary; ask your brain for “that article about pricing” months later.

🗣️

Voice dictation

Hold to talk — live, Wispr-style, with an AI clean-up pass. Speak anywhere you’d type.

🎙️

Meeting recorder

Record a meeting → transcript → a clean summary with takeaways and decisions, kept in your brain.

🧩

Claude Skills

Reusable skills you’ve built — your brain knows them and can run them on command.

the standout

💡 Ideas → an agentic workflow

Drop in a rough idea and your brain grows it into a real plan — then hands you a ready-to-run agentic prompt for a coding agent. From a sentence on the bus to a workflow an AI can execute. Your ideas don’t just sit there; they get built.

Chapter 12 · the engine room

You run the AI

Most “AI apps” quietly run up an API bill and hide the controls. My Brain hands them to you — pick the engine for every feature, read and edit the prompts, watch the usage, and run the heavy work on subscriptions you already pay for.

Near-zero API cost

Background AI work runs through your Codex and Gemini CLI agents — on subscriptions you already have — instead of metered API calls. If an engine is ever down, it falls back automatically to Claude Sonnet so nothing stalls. Real power, without the runaway bill.

⚡ Codex CLI ✨ Gemini CLI 🛟 Sonnet fallback
🎛️

A model picker per feature

Choose which engine writes your story, your email brief, your commitments — feature by feature, in Settings.

✍️

Edit the prompts

The system prompts behind every feature are yours to read and tune. Nothing is a black box.

📊

See the usage

A usage view shows what the AI is doing under the hood — no surprises.

🗂️

Control the memory

The Index manager: turn any section of your memory on or off, see RAG-vs-cloud counts, and Reconcile in one tap.

In short: you are in control of the AI, the models, the cost and the memory — not the other way round.

Chapter 13 · the full catalog

Everything it does

The whole brain, end to end — nothing left out. Sixteen areas, a hundred and fifty-plus little powers.

Chapter 14 · the road ahead

It’s only getting started

The brain lives in your pocket and on your server today. Next, it reaches into the physical world.

🤖

Hardware

Dedicated devices for a few focused use-cases — so capturing a thought and getting it back is as natural as speaking into the air.

in the works
📱

A native Android app

A purpose-built Android app for a handful of high-value use-cases — faster capture, deeper notifications, always within reach.

next up

Same brain, more doors into it.

Chapter 15 · built right

Built by AI — held to discipline

Every single feature shipped through an enforced pipeline. Nothing reaches you on trust.

🎫
Ticket
every feature tracked
⌨️
Code
one branch per feature
🧪
Tests
red = nothing ships
🚀
Deploy
all-or-nothing
Verified live
then — only then — done

Versioned privately · data pulled to a second machine at home every night · every backup integrity-tested · a watchdog guarding it all.

From zero to a life companion — in days, not years

It was born

Login, the twin-memory engine, the design system.

It started remembering

Capture, bookmarks watched by AI, ideas and skills organized.

It started listening

The daily loop, the Story of the Day, Telegram by voice.

It started understanding

The mentor, focus areas, on-track score, suggested tomorrows.

It connected to your world

Gmail, meetings, commitments, connections, semantic Explore.

It learned to keep secrets

A zero-knowledge vault — and next, hardware and a native app.

Most apps manage what you do.
This one understands who you are.

Every day it runs, it knows one day more. The version you meet next year will know its owner better than any product ever has.

🧠 Open My Brain →