Your AI knows YOU.
Every answer cites the sources you brought. Your journals, your surveys, your context — your AI reasons over them. Not the internet. Not yesterday's headlines. You.
Personal Context IntelligenceEarly access
Personal Context Intelligence.
The AI that finally remembers you.
Big Tech's AI forgets you between sessions. Ours doesn't. Bring your sources, write your journals, take a few surveys — your AI catches up. Permanently. Grounded in your own context, never the internet's.
Private by default. Citations on every claim. Export anytime, in plain markdown. No ads. No data sale.
Free to try. $12/month to keep. No credit card to sign up.
Private by default. Mathematically.
Row-level isolation. Encrypted at rest and in transit. No model training on your work. No ad-targeting. No data sale. The architecture is the privacy claim.
Export anytime. We earn trust by making leaving easy.
One click. Plain markdown. Everything you wrote, everything you cited, everything you taught your AI. Yours, the day you arrive and the day you leave.
How it works
Four steps. Ten minutes. One AI that knows you.
The smallest commitment we can ask is one source and one question. From there, the personal-context layer compounds every time you write, search, or take a Survey.
- Step 01
Bring your sources.
PDFs, URLs, paste. The notebook attaches them. Your AI indexes them so every answer can cite the page it came from.
- Step 02
Ask a grounded question.
Open the chat. Ask. Every claim in the answer cites the source page it came from. If support is weak, your AI says so — no fake confidence.
- Step 03
Write a journal entry. Take a Survey.
Two minutes of writing. Three minutes of answering. Your AI now has durable context it can reason with — only when you say so.
- Step 04
Your AI catches up. Permanently.
The next time you open Genie mode, your full personal context is on tap. Inspect every excerpt. Export the whole account in plain markdown, one click.
What a session looks like
Every claim cites the source. Every excerpt is inspectable.
No mystery context. The right-hand pane shows the actual passages your AI consulted. If the support is weak, the answer says so — instead of inventing confidence. Tap a citation below to see the excerpt it came from.
Quarterly review · notebook
IllustrationNotebook chat
You
Across the two quarterly reports, which segment grew fastest — and what drove it?
Journal Genie
Cloud infrastructure grew fastest— about 34% year over year. The driver was one enterprise contract renewed across the quarters, not broad new-logo growth. Support for a “broad new demand” reading is weak in these two sources, so I'm flagging it rather than asserting it.
Sources
- Q2-earnings.pdf
- Q3-earnings.pdf
Cited excerpts
[1] Q2-earnings.pdf · p.14
“Cloud Infrastructure revenue grew 36% YoY in Q2, outpacing every other reporting segment.”
[2] Q3-earnings.pdf · p.22
“The Acme Corp renewal accounted for the majority of segment growth; net-new-logo contribution was ~8%.”
Saved to Studio · “Segment growth — Q2–Q3 synthesis”
An illustrative walkthrough. In the product, every citation links to the exact page in the source you uploaded.
Six surfaces. One AI that knows you.
The product is the answer to “how do you know I won't forget my own thinking?”
Notebook
Bring your sources. Ask grounded questions. Every claim cites the page it came from. No black-box AI.
Genie
Your full personal context — notebooks, journals, completed surveys — on demand. Inspect every excerpt.
Journal
Write privately. Your AI reads it tomorrow — only if you say so. Permanent context, not surveillance.
Surveys
Quick Polls and Deep Surveys teach your AI what you actually think. Two minutes; permanent memory.
Studio
Save useful outputs as durable artifacts. Editable. Re-citable. Built for thinking that compounds.
Export
Plain markdown. One click. Same export the day you arrive and the day you leave. Yours, forever.
The closest comparison
Journal Genie vs. NotebookLM
NotebookLM is Google's excellent source-grounded notebook — and the closest thing to Journal Genie's workspace. Source grounding and citations are its category-defining strengths; we don't pretend otherwise. Here's where we deliberately differ: a private journal and surveys that compound your context, an export that lets you leave, and a business model that isn't funded by your attention.
| Capability | Journal Genie | NotebookLM |
|---|---|---|
| Grounded answers that cite the source passage | Yes. Core to the Notebook surface | Yes. The category-defining feature |
| A private journal that compounds your personal context | Yes. Private by default; opt-in to AI per entry | No. Source-only — no journal surface |
| Structured surveys + polls that teach your AI what you think | Yes. Quick Polls and Deep Surveys | No. Not a feature |
| Full-fidelity export of your whole account, in plain markdown | Yes. One click. Part of the trust contract. | No. No built-in full-account markdown export |
| Works without a Google account | Yes. Email sign-up; Google sign-in optional | No. Requires a Google account |
| Funded by you — no ads, no data sale, no training on your work | Yes. Twelve dollars a month. That is the whole model. | Part of Google’s ad-funded ecosystem |
| Per-user database isolation, tested adversarially | Yes. 30/30 row-level-security checks pass | Not publicly documented |
Comparison reflects publicly available information about NotebookLM as of May 2026. NotebookLM is a Google product; “NotebookLM” is a trademark of Google LLC, used here only for identification and honest comparison. If any row stops being accurate, we update it within 7 days — the same standard we hold our own claims to.
What Big Tech does that we don't
The contrarian frame, with receipts.
We earn the right to be loud about privacy by being specific. Every claim below is currently true, substantiated by a public policy or product behavior you can verify yourself.
Big Tech
Google's Gemini privacy policy permits training on your conversations to improve services and personalize ads.
Journal Genie
We do not train AI on your journals, sources, or chats. Period.
Big Tech
ChatGPT forgets you between sessions. Every conversation starts cold.
Journal Genie
Journal Genie remembers your sources, journals, and surveys — permanently, until you delete them.
Big Tech
Big Tech AI products are funded by ad-targeting, attention capture, or training-data extraction.
Journal Genie
Journal Genie is funded by you. Twelve dollars a month. No ads. No data sale. No surveillance.
Big Tech
Most "AI productivity" tools wrap the same OpenAI API everyone else has — with no personal context layer.
Journal Genie
Personal Context Intelligence is the layer they don't have. Your sources, your journals, your surveys — grounded, citable, exportable.
Built in public
We're pre-launch. Here's what's real today.
No fake testimonials. No "trusted by thousands" theater. The numbers below are facts you can audit from the public repo or by stepping through the test suite yourself. When the first real customer quotes arrive, they will live in this same slot.
dev = prod
Migration parity, dev + production Supabase
The production database is migration-current with the development project. Schema parity is verified after every release. The architecture is real, not a Figma mock.
3,700+
Vitest unit + integration tests, all passing
Every hardening sweep adds tests, not anecdotes. Property-based tests where the input space is large. Mutation-aware where it counts.
30/30
Adversarial row-level-security checks PASS
Cross-account read attempts blocked at the database layer. Not an audit promise — a query you can't complete from the wrong session.
0
Production security-advisor warnings
Supabase Security Advisor reports zero WARNs on the production project. The next number on this card will be a CVE we found and shipped a fix for, not a badge we bought.
What people say
No testimonials yet — and we won’t invent them.
We're pre-launch. There are no customer quotes here yet — and we won't fabricate them. When real users say something worth quoting, with their explicit permission, it will live in this exact slot. Until then, the proof is the architecture: the numbers on this page are auditable from the public repo and the running test suite.
Start free
Free is the whole product — at a lower ceiling.
Not a feature-gated trial. The Free plan runs the same code path as paid — every surface, real AI credits, full export. The day you sign up, your AI starts knowing you.
- 40 AI credits every month
- 10 source imports every month
- All six surfaces — Notebook to Export
- One-click, full-account markdown export
- Up to 3 notebooks · 20 stored sources
- No credit card. No ads. No data sale.
Outgrow it and Pro is $12/month, Premium $24 — annual billing is ~20% less. Cancel anytime, end-of-period. No retention dark patterns.
Where we're early
Honesty isn't the opposite of bold. It's the proof.
- Early access. We say so on every page. No "trusted by millions" theater.
- Pricing matches the code that exists today. The Pro and Premium ceilings are real and server-enforced.
- Counsel review is in flight on the legal-trust pages. Where text is provisional, we mark it.
- GENIE on-chain claims open after the distributor deploys on Base. Earning points works today — claiming as tokens waits for deployment evidence.
Journal Genie is in early access. The core product is materially landed; broad public-launch hardening continues. We say so on every page so you can decide for yourself.
Start the way we want to be judged
Bring one source. Ask one question. Decide if your AI should know you better.
That's the smallest commitment we can ask. If grounded answers from your own sources beat a chat that forgets you tomorrow, the rest of the product follows naturally. If not, export everything and walk away. We mean that.