Private Beta
Found a bug? Tell us

Hand over
the operation.

The mail read, the setlists reconciled, tomorrow's brief waiting when you look up. Albria is your COO — she runs your work and your life across every device. You direct. She executes. Powered by your own Claude account.

6:47 AM needs you

Found the insurance appeal deadline buried in Thursday's mail. The letter is drafted — one decision is yours.

2:15 PM

While you were in rehearsal: setlists from all three tapers reconciled, the master updated.

9:30 PM

Tomorrow's brief is ready: two things need you, she handled the other eleven.

Native on MaciPadiPhoneWatch

Scroll to see what she is
Albria macOS app — the real thing, running

The actual app, on a Mac — not a render.

In sixty seconds

What it is

Your COO — one operator that runs your work and your life, across every Apple device.

What it does

Detects, organizes, and executes — then surfaces only what needs you.

Plans your day Ships your code Runs errands Books your travel Learns with you Never drops an idea

What you need

Claude account Tailscale API keys GitHub
Why these →

Free on Mac · $1/mo on iPhone, iPad & Watch. You pay Claude, GitHub, and Tailscale directly — no markup from us. See full breakdown →

Ships daily. New builds land every 24 hours — if something's wrong, it may already be fixed in today's build. Get the latest →

Orientation

What you're looking at

This is Albria — a native Mac app where one operator runs your work and your life. One window, four surfaces. Here's the actual screen, and what each part is for.

Albria on a Mac — Focus Board on the left, her chat below it, a report she produced in the center

Focus Board

The threads of your work and life. A champagne dot means one needs you.

Tasks & Monitor

Watch her actually working, step by step.

Chat

You talk — type or voice. Fire every thought.

Briefs

What she brings you in the morning — and after every run.

Each screen answers one question you have: What's happening? What needs me? Let me say something.

How it works

How a thought becomes work.

Plain mechanics, no magic — the same four beats whether it's a song, a trip, a case file, or a business.

You say it

Typed or spoken — half-formed is fine. Say it the way you'd say it to a person.

It becomes a thread

She shapes it into a piece of work, and you'll see it appear on the board.

She executes

With your real accounts and tools — your Claude account, your files and folders (drafts, songs, case documents, spreadsheets), your calendar, a browser she can drive. And for the technical: your code.

It lands back

Finished — or wearing a champagne dot when it needs your judgment.

Nothing hidden in the middle: open Monitor and watch every step as she takes it.

Fair questions

The questions you should be asking.

How is this different from ChatGPT?

A chat answers you, then forgets. She works for you — in your real files, your calendar, your projects — all day, remembering everything, and brings back finished things for your judgment. A chat is a conversation; this is an operation.

What actually is this?

A native Mac app where one AI operator runs your work and your life — not a chatbot in a browser tab.

Where are my files?

Where they always were — your folders, on your Mac. She works in them; nothing moves into a silo.

How does the AI part work?

You sign in with your Claude account — the same one behind the Claude chat you may already use — and she puts it to work. Every step she takes is visible in Monitor; you can watch her work, not just trust it.

Is there a browser?

There's one built in. She can browse alongside you, or drive it herself when a task needs the web — booking a table, chasing down a source, filling a form — and you watch it happen.

Do I have to manage her?

No. You direct and decide; she reports back. A champagne dot means a decision is yours — silence means it's handled.

Will this feel strange?

For about a week, yes. You'll want to check everything she does — that's a lifetime of habit, not a product flaw; watch her work in Monitor until trust arrives on its own schedule. The strange part isn't learning the app — it's discovering what your attention is for once the busywork isn't yours.

Letting go is a practice, not a switch. The long version of that answer lives in the philosophy →

Think out loud. She catches all of it.

Albria absorbs ideas as fast as they come — in any order, across unrelated domains — and nothing drops. You don't have to slow down, queue your thoughts, or finish one before starting the next. Fire a calendar change, a product idea, a bug, a song you want to learn, a new app — she catches every one, holds the thread, and acts.

If your ideas have always come faster than any tool could hold them — this is the one you've been waiting for.

You, firing
"Move my 2pm to Thursday." caught
"Add a dark-mode toggle to the app." building
"Prod's throwing 500s on /users." caught
"Teach me 'Ripple' on guitar." planned
"Idea: a newsletter for the launch." caught
"Also — build me a show-finder app." on it

Six unrelated thoughts. Zero dropped. She's already holding the thread on all of them.

One Albria. Everything you run.

Work, life, and the things in between — directed by you, run by one operator. A few of the things people hand over:

Run your life.

Calendar, errands, reservations, travel — handled, not just tracked.

You: Book us a table Friday and sort the SF trip.

Albria: Done — 7:30 for four on OpenTable, flights booked, and a hotel near the venue. All three on your calendar.

feel this one →

Learn an instrument.

Build a song on guitar, one practice at a time.

You: I want to play 'Ripple' on guitar.

Albria: Pulled the chords, broke it into a 2-week practice plan, 15 minutes a day on your calendar. I'll track your progress.

feel this one →

Get in shape.

Workouts and diet that actually adapt.

You: (nothing)

Albria: Noticed an empty training week — here's a 3-day plan around your schedule. Booked it. I'll adjust when life happens.

feel this one →

Run a business.

Your COO for the whole operation.

You: How are we doing this week?

Albria: Triaged every channel. 40 things handled, 2 decisions only you can make — here they are.

feel this one →

Ship an app.

Idea to App Store — Albria builds it.

You: Build me a Grateful Dead show-finder.

Albria: A fleet of agents wrote it, tested it, and shipped it overnight. Apps like DeadDonut and BillBattle were built exactly this way.

feel this one →

Just have company.

Companionship and presence — the things that aren't tasks.

You: Put on something calming with Mom.

Albria: Queued a musical you both love. I'll keep the night easy — no work talk.

feel this one →

...and an infinite number more — limited only by your imagination.

These six are where people start. There is no end to the list — and none of it is six separate apps. It's one surface that already knows the context for all of them.

One place for your whole life.

Your life is one thing, but your tools treat it as twenty. The calendar app doesn't know the deadline lives in your repo. The repo doesn't know Thursday is Mom's appointment. Nothing knows the band plays Saturday.

Albria is the one surface that knows all of it — the work deadline, the appointment, the gig — because she's the one running all of it. So "what does my week actually look like?" finally has a single, honest answer.

Not another app to check. The place where it all comes together, quietly.

Albria on iPad — one surface for work and life

The real app on iPad — everything above happens in here.

Thursday — one calendar, one life

work · family · music — one surface

Auth PR ships — work deadline, tracked from the repo
Mom's appointment, 2pm — leave by 1:40, she's already watching traffic
Band rehearsal, 7pm — setlist notes ready from last week

Three domains that usually live in three different apps — one view, with the context already attached. The champagne dot is the one that needs you; everything else is already handled.

What a day looks like

Six moments from a real day — directed by you, run by Albria.

6:30 AM Wake Up Brief
Albria
Good morning. 3 priorities today: ship the auth PR, dentist at 2pm, Mom's birthday gift.

Your day is planned before your feet hit the floor.

9:30 AM Bug Triage
You
We got a Slack alert — prod is throwing 500s on /api/users.
Albria
Found it. The migration from last night left a null constraint on users.display_name. I've opened a PR with the fix and a backfill migration.

Root cause identified and PR opened in under 3 minutes.

11:00 AM During Meeting
Albria
Meeting notes captured. 4 action items extracted:
1. Update nav component (you, by Friday)
2. Review color tokens (Sarah, by Wed)
3. Schedule user testing (PM team)
4. Follow-up meeting Thursday 2pm — added to your calendar.

Action items captured and scheduled automatically during the meeting.

5:30 PM Life Admin
You
Order Mom's birthday gift — she likes gardening and mystery novels.
Albria
Found three options: a Louise Penny boxed set ($32), a Japanese hand pruner set ($45), or both bundled with gift wrap ($72). Want me to order the bundle?

Gift researched, selected, and ordered in 60 seconds.

7:00 PM Side Project
You
Let's work on the recipe app for 30 minutes.
Albria
Switching to Recipe Lab. Last session: you added the ingredient parser. Next up was the unit conversion logic. Timer set — I'll let you know at 7:30.

Clean context switch to personal project — no setup, no friction.

9:00 PM Wind Down
You
How'd today go?
Albria
Solid day. Auth PR shipped, prod bug squashed, dentist done, Mom's gift ordered. Tomorrow: nav component, code review responses, Thursday standup prep.

Daily summary with tomorrow's plan — no loose ends.

How you drive it

You direct. She runs it.

You don't hand over your life on day one. You hand over one small thing — a calendar move, a bug, a gift that needs ordering — watch it come back done, and hand over a little more. Letting go is a practice, not a switch: you run your own experiments, at your own pace, and the authority she holds is the authority she has earned.

The arc is real, though. This project started with driving Albria like a pair-programming engineer — close to the code, reviewing each line. Each month of compounding trust moved the work up the ladder: less implementation, more direction. Today you talk to Albria the way a COO talks to their organization — in intent and judgment, not instructions.

01

Start with one experiment

Pick one thing you'd normally do yourself and hand it over. Verify the result. That loop — delegate, verify, expand — is the whole on-ramp, and you control its speed.

02

Fire every thought

Every random idea, task, worry, or half-formed plan — you just say it, out loud or in passing. She decomposes it, files it, tracks it, and comes back only when she needs a decision. Nothing lands in a black hole.

03

Meet on strategy

Like any executive team, you meet. Scheduled sessions to think through direction, security, features, products — the calls that deserve real deliberation. You set the where and why; she carries the how.

What to Expect

  • Speak in intent, not instructions. Describe the outcome; she figures out the steps.
  • Fire ideas at the speed of thought. Any domain — work, life, a new product — nothing gets dropped.
  • Silence is healthy. She surfaces only what needs you — a decision, a risk, a result.
  • Leverage grows with trust. The more you hand over and verify, the more she runs on her own.

The honest part: this doesn't hand you back empty afternoons. It moves your time up the ladder — out of execution and into the judgment, taste, and direction only you can give. That trade is the whole point.

Why can you keep handing over more? Because of one promise the data-brokers can't make: your life stays sovereign. Local-first, on your own accounts and devices, never sold, never mined, never used against you. You only hand someone more of your world if that world stays yours — and that's not a feature we added. That's the architecture.

Hand over as much as you're ready to. Then watch how fast "ready" grows.

Want the maximal version of this philosophy — full-send, and why we built for it? How to think about using Albria →

Don't read about it — feel it click.

There's an interactive demo on the getting-started page — type a thought, watch it become something you can act on. Two minutes, in your browser, no install.

Try the demo

Start with one Mac. Scale to a 24/7 operation.

You don't need a server farm to begin — just a laptop and your Claude account. Grow your capacity only as far as you want to trust it.

1
Download + connect Claude

One Mac.

Download Albria, connect your Claude account. Your COO is live on your MacBook — free. That's the whole setup.

2
Add devices to your mesh

All your devices.

Add your iPhone, iPad, and Apple Watch to a private Tailscale mesh. Now it's one Albria everywhere you are — direct it from your wrist, pick it up at your desk.

3
Add machines — yours or Alembic

A fleet that never sleeps.

Add machines — bring your own, use any provider, or spin them up from Alembic Compute, which is purpose-built to drop straight into a Albria fleet (close to one-touch, no hardware to own). Alembic's the easiest path — never the only one. Now a fleet of agents works around the clock while you sleep.

1 Mac
~6 Macs
50+ Macs

You grow from one laptop to a fleet of n — at your pace. Capacity scales with trust, not commitment. Each step is a single, clear action.

Start free on one Mac. Add machines only when you're ready to trust more.

What you connect — and why it matters.

Albria is the conductor, not the orchestra. The intelligence, the network, the voice, the code — you own all of it, directly, with no middleman. Here's exactly what you connect, what each does, and why we deliberately don't bury it inside our app.

Never heard of some of these? Most people haven't. Albria walks you through each one once — in plain language, set up for you — and then you never think about them again. We've abstracted the technical parts so anyone can use them.

Claude account — the intelligence.

What it is — Your Claude subscription (Claude Max) or API key — the brain Albria thinks with.

Why you bring it — Albria conducts; Claude reasons. You pay Anthropic directly — we never mark up a token or meter your usage. Your account, your terms. No resale, no margin, no lock-in.

Claude is the brain Albria runs on today — primary and recommended, the model she's tuned for. But it's your brain: she already runs on other model CLIs too, and choosing yours is part of the deal. More below →

Tailscale — the private connection.

What it is — The private, secure connection between your own devices — your Mac, your phone, your machines — so Albria reaches them directly and no one else can.

Why you bring it — It's your network identity, not ours. Your traffic never routes through our servers, because we never want to be able to see it. Privacy by architecture, not by promise.

API keys — optional superpowers.

What it is — One key unlocks one service. Bring your own for anything you already use and Albria operates it: voice (ElevenLabs), social (X, Instagram, Facebook), markets (Alpaca, Kraken), reservations (OpenTable), media (Archive.org) — and anything else with an API.

Why you bring it — You hold the key, you control the spend, you revoke it in one click. No bundled markup, no surprise bill from us — just the services you choose, at cost.

The idea — The API key is the universal unlock. Albria isn't a fixed list of features — it's a conductor that runs whatever you connect, across every corner of your life and work.

GitHub — a history of everything you make.

What it is — A complete, private, recoverable history of your work — your novel, your songs, your business plans, your code. Not just for developers: Albria turns version control into something anyone can use, so nothing you make is ever lost and every version is recoverable.

Why you bring it — Your account, your permissions, your history. Every change is attributable to you and revocable by you — we never hold your work hostage or sit between you and it.

Your brain, your choice.

Albria conducts; the model reasons. Which model is the brain is up to you — choosing it is part of owning your substrate, not a setting we hide.

Today

Claude is the default and what Albria is tuned for — primary and recommended. But she already runs on other model CLIs too: we're driving real work through Codex right now. You bring the account; you pick the brain.

On the roadmap

Swap Albria's brain between Claude, Codex, Gemini, or a local LLM — your operator, your model. Full brain-swap (especially Gemini and local) is on the way, not a shipped matrix yet.

What it costs — all of it, up front.

No surprises. Because you bring your own accounts, you pay those providers directly — Albria never marks up a cent or meters your usage.

Albria on Mac

Free

Forever. No trial, no catch.

Albria on iPhone · iPad · Watch

$1/month

App Store. Cancel anytime.

AI is the new utility — cancel the cable bill and build something instead.

The costs that aren't us — you pay these directly.

Claude — the brain.

Your main cost. Albria runs on your Claude subscription — you pay Anthropic directly.

Free / Pro Free – ~$20/mo
Max 5× ~$100/mo
Max 20× ~$200/mo

More accounts = more parallel capacity. Prices are approximate — check Anthropic directly.

GitHub — free for you.

Unlimited private repos on the free plan. Individual accounts get everything they need at no cost.

Personal (what you need) Free
Team features ~$4/mo

Paid tier only if you need multi-user team management — you won't need it to start.

Tailscale — free for you.

Personal use is free. The private network that connects your devices costs nothing for individuals.

Personal (what you need) Free
Business / team scale Paid

Paid tier only at team/business scale — individual Albria use is free.

What it costs as you scale.

Same Albria at every rung. Your costs are just the third-party services you use — and only the ones you turn on.

Test it. 1 Mac

Albria free on Mac + Claude free tier + free GitHub + free Tailscale. Build a real proof-of-concept without committing real spend.

Free

basically

Ramp up. ~6 Macs

Claude Max 5× + the free tiers of everything else. One person, daily use, real output.

~$100/mo

total est.

Run hard. ~6+ Macs

Claude Max 20× — fleet-scale output for a solo operator running at full capacity, 24/7. One account caps out fast at this pace; stack a second or third Max account for more parallel throughput.

$200–$600/mo

total est.

Run companies. 50+ Macs

Multiple Claude accounts + compute (your Macs or Alembic Compute) + team tiers as you need them. No ceiling we impose — your cost is purely proportional to your capacity.

You decide

no ceiling from us

You'll never get a surprise bill from us — there's nothing for us to surprise you with. You see every cost before you commit to it.

Ready to
hand it over?

macOS app is free. iOS coming to the App Store.

Currently in private beta. Public beta in 3-4 weeks. Independent product by Graham Alembic LLC.

No spam, ever. Unsubscribe anytime.

A Note on Reliability

Be clear-eyed about what this is: Albria's current workflow runs on the Claude command-line tool. That is the engine today. When Claude's service is degraded or down, Albria is degraded or down with it — there is no separate cloud we keep running in its place. Multi-model support (Codex, Gemini, and on-device models) is on the roadmap and will soften this, but right now the dependency is real and singular.

This technology is maturing in real time. Providers ship breaking changes, move rate limits, and larger forces are in play — a release pulled, a frontier model restricted by a government, an account throttled without warning. Any of these can take the app offline for a stretch, through no fault of your own setup.

So expect some disruption. The AI providers are nowhere near the "five nines" (99.999%) uptime you'd assume from mature infrastructure — they are research labs and startups operating at the frontier, not utilities. We build for resilience and we keep widening the set of models you can fall back to, but we won't pretend the ground is steadier than it is.

This is the same honesty we bring to everything: name the risk plainly, then let you decide. The risks are real. The rewards are real.

Read the full philosophy