Remembers your life. Keeps it yours.

allMe checks in with you daily, writes your week, and answers questions like “which restaurant did we go to on Mum’s 60th?” — searching everything you've written, said and photographed, all of it on your phone.

See how it works

iPhone · Pre-MVP, building now · Free tier, then $59.99/year

487 memories kept

Today

Wed 19 Aug

How was today?

From last week

You were nervous about the meeting with Sarah. How did it go?

Answer

Not now

Tomorrow · 6pm

Text Marcus about dinner

You said you'd sort it before he leaves for Berlin.

Remind me

Not needed

Two years ago

The last night in Lisbon

19 Aug 2024 · 6 photos

487 memories kept this year

Today

Ask

The product

Three seconds to keep something.
One question to find it again.

You never choose “note or diary” — allMe files each capture and shows a chip you can flip. Voice, photos and text all land in one inbox.

Ask

Which restaurant did we go to on Mum’s 60th?

Vico, on the corner by the church. Mum cried when the waiters sang.

14 March 2024. You booked it six weeks ahead — the only Saturday everyone could make.

14 Mar 2024

1 entry

Try asking

When did I last see the dentist?

What did I promise Sarah?

Ask your life anything

Ask Answers come from your own entries, with the entry attached.

Cancel

0:38

Ran into Marcus at the market. He's moving to Berlin in the spring — we said we'd get dinner before he goes.

Diary

Change

Transcribed on this iPhone

Capture Speak. It transcribes on the device and files it for you.

Sunday story

Your week

11 – 17 August

memories

photos

voice notes

A good week, mostly. Two wins at work and one hard Tuesday you didn't want to talk about. Marcus came up twice.

One thing I noticed

Every good day this month started outdoors.

Sunday Story Your week, written for you — the payoff that keeps people coming back.

A check-in that remembers

It follows up on what you said last week. That continuity is the moment a trial becomes a habit.

Answers nothing else has

“What did I promise Sarah I'd do?” Your diary knows. Your email doesn't.

Notes live here too

Recipes, wifi passwords, quotes. One inbox, sorted automatically — no filing.

Private by construction

Entries and the search index never leave your phone. We can't read your diary — we don't have it.

Why now

2026 is the first year the private version is buildable.

Apple now gives third-party apps free on-device foundation models with image understanding, free best-in-class on-device speech-to-text, and free privacy-preserving cloud compute for small apps. Small embedding models and mobile vector search matured in the same window.

“We can't read your diary” becomes architecture

Not a policy a subpoena or an acquisition can undo. The entries and the AI's search index both stay on the device.

The cost structure inverts

Our marginal AI cost is near zero while incumbents pay per token for every user — so we can run the category's most generous free tier and still profit.

The market

Every competitor is AI-native or actually private. None is both.

Digital journaling is a $6–7B category growing about 11% a year. The revealing part is who hasn't won it.

Positions are strategic judgements, not measurements. The point is the shape.

Who	Where they are in 2026	How we win
Apple Journal free	Solid basics, preinstalled. No chat, no cross-entry synthesis, no notes fusion.	Don't fight the basics — win on memory and synthesis, and out-ship an annual release cycle.
Google Gemini free	Searches your Gmail, Photos and Drive — your data exhaust.	They never hold the self-authored narrative: what you felt, decided, promised yourself.
Day One $75/yr	AI bolted onto a 15-year-old journal; encryption has documented gaps.	AI-native from day one, lower price, a genuinely on-device index.
Rosebud $108/yr	AI-native leader, ~7,500 payers. Cloud-stored; mid-entry paywalls users resent.	Memory over coaching, never interrupt an entry, provable privacy.
Finch $30M ARR	A self-care pet, bootstrapped, no AI at all — the retention champion.	Not a rival, a lesson: emotional stakes beat feature count.
allMe $60/yr	AI-native and on-device: daily check-in, weekly stories, ask-your-life search.	The empty quadrant, at a fifth of what the only private competitor charges.

Privacy

The promise is a build constraint, not a marketing line.

All content and all embeddings live on the device. The only things that ever leave: encrypted backups to your own iCloud, and — when you ask a question — the few excerpts needed to answer it, sent to a model that doesn't retain them.

Four privacy levels per entry. Normal, hidden from AI, this-iPhone-only, or locked behind Face ID.
No accounts, no ads, no data sales. Zero tracking SDKs, now or ever.
Export everything, always free. We never hold your memories hostage.

Why the index has to be local: published research recovers roughly 92% of short texts exactly from their AI embeddings. “We only store vectors” is effectively “we store your diary” — so ours never leave the phone.

Privacy Level

Set it per entry. Change it any time.

Normal

Backed up · searchable

Hidden from AI

Never searched by AI

This iPhone only

Don't sync with my other devices

Locked

Face ID to open

Business model

Free where it builds trust. Paid where it creates depth.

Free, forever

Unlimited capture, on-device search, memories resurfaced, every privacy feature, full export. A paywall never interrupts an entry in progress.

$6.99/mo · $59.99/yr

Unlimited memory chat, Sunday Stories and monthly letters, semantic photo search, encrypted multi-device sync.

Later: things people gift

Shared journals for couples and families — which recruit their second user by design — and printed year-in-review books.

Worst-case AI cost runs about 10–15% of revenue, and realistically far lower because on-device models absorb most of the work. Embeddings cost nothing, forever. No ads, no data sales — the trust is the business.

The Foundry

What we bring to the eight weeks — and what we intend to leave with.

The bootcamp runs problem clarity → market mapping → customer discovery → traction → pitch. Each stage already has work behind it or scheduled for it. We arrive with a thesis; we should leave with evidence.

Stage	What's already here	What the eight weeks add
Problem clarity	A specific, evidenced problem — retention data, and a court production that makes the trust risk concrete.	Confirmation from people who actually quit journaling apps.
Market mapping	A researched competitive map with a demonstrably empty quadrant.	Pricing sensitivity and switching-cost tests against Day One and Rosebud.
Customer discovery	Four personas drawn from category research, weighted to lapsed journalers.	20+ interviews, with the lapsed segment as the primary target.
Traction	A working architecture and a 16-week build plan already costed.	Landing page, 300 waitlist signups, and a TestFlight beta in real hands.
Pitch	One-liner, why-now, moat and business model articulated.	A pitch carrying a retention number instead of a claim.

The eight-week plan

Weeks 1–2

Customer discovery20+ interviews: current journalers, lapsed journalers, new parents, privacy-conscious professionals. The lapsed segment knows exactly why these apps fail.

Weeks 3–4

Demand signalLanding page live and 300 waitlist signups, testing three positioning lines: privacy, memory, daily ritual.

Weeks 5–6

Working proofA TestFlight build of the core loop — capture, index on device, ask your own life a question.

Weeks 7–8

Evidence30–50 beta users. Measure day-seven retention and check-in completion; test willingness to pay at $59.99/yr.

The one number we intend to bring to Catalyst: day‑seven retention against a category baseline of 8%.

Beat it and the thesis is proven. Miss it after two iterations and the give-back thesis is wrong — which is worth knowing, and is exactly the kind of answer eight weeks should produce.

Questions we expect, answered short

Apple ships this next year — then what? Apple ships basics annually; we ship the memory loop weekly. Their corpus is your data exhaust; ours is your narrative.
Isn't this a feature, not a company? The artifact is the company: an owned, exportable life archive with a subscription and, later, printed books. Companions get abandoned; archives compound.
Why will you retain users when 87% quit? Because we invert the give/get ratio from day one — and we'll have the number to show for it.
Isn't on-device AI worse? For classification and dialogue it's now sufficient; heavy synthesis routes to privacy-preserving cloud with only the excerpts needed.

Stage and fit

Pre-MVP. Product strategy, competitive research and technical architecture are complete and written down; the build starts now. Industry: Software, SaaS & AI-enabled Products.

Deliberately not positioned as health or therapy — four US states restrict AI-therapy marketing, and allMe is a memory and reflection tool, not treatment.

Founder
Leonard Eshun — [ONE LINE ON YOUR BACKGROUND AND WHY YOU CAN BUILD THIS]

[WHY THIS PROBLEM IS PERSONAL: the moment you tried to remember something and couldn't]