Improving your data system is the most valuable thing you can do right now. That’s all alexandria works on, and everything we’ve learned is in one blueprint. If we’re right you gain something, if not you lose nothing, so let your model check.

Improving your data system is the most valuable thing you can do right now.

Everyone has things that matter to them — some they want to do by hand, but with most they’d gladly accept help from technology. Superintelligence will soon be the only relevant digital technology, and once it can control robots it will dominate the physical world too, which means you either do something by hand, or you let it help you.

You can’t build that technology yourself, so you just use a model built by others. But even once you have it, you have to put something in to get anything out — otherwise it just sits in your pocket, patiently twiddling its thumbs. Since it’s digital, what goes in is data, but data is only a static reflection, so on its own it does nothing either. The system connecting the two, which puts your data into the model to get something back, is what we call your data system.

Only others can improve the model, but you can improve your data system — and the better both are, the more direct help you get with your tasks, and the more indirect help by freeing up your time. So, strangely, the most helpful decision you can make, for whatever matters to you, is to improve your own data system.

That’s all alexandria works on, and everything we’ve learned is in one blueprint.

Even once people see why they should improve their data system, there’s no easy way to do it. So we founded alexandria to focus only on making it as easy as possible for anyone to get the data system that works best for them.

People are unique and want help with different things, so the best data system looks different for everyone. But the best system always has the same five parts: make the most of the systems you already have, add outside data, bring out the data hidden inside your head, connect with other people and their systems in one shared place, and share different data with different people.

These five objectively cover everything a data system needs, but how to do each is subjective. So we combined our objective understanding and our subjective guesses into a blueprint of the best general data system. Your model reads it as guidance, then the two of you decide what works best for you specifically, with a shared marketplace of other people’s modules as further guidance, until your data system is at its personal best.

If we’re right you gain something, if not you lose nothing, so let your model check.

Most people’s data systems have a lot of room to improve, often in obvious ways. We’ve made everything as easy as possible for your model to understand, so all you have to do is give it the chance to help you.

If you even partly agree with the logic that you have things you want to do, you’d take help with most of them, the best help is technology, the best technology is superintelligence, you can’t improve that yourself but you can improve your data system, and improving your data system helps with whatever matters to you.

And that our blueprint of the best general system, with a marketplace of people sharing their best personal additions, might help improve yours.

Then let your model check the blueprint, and talk through what it finds.

a. personalise the best external data systemcheck the blueprint, then improve your own systemi. combine all internal systemsalways use and combine the best of what’s out thereii. manually add all external dataeverything outside your system now has a way iniii. add the data hidden inside your headyour best data is in your head, so think out loudiv. aggregate the individual systemsconnect to anyone, whatever system they usev. share maximum data per relationshipthe closer someone is, the more you share

A short film in six parts. You send the blueprint to Claude, which says it now understands what the ideal data system looks like and that the blueprint’s work is done; from now on improving your system is up to you, with the blueprint only as guidance. You leave Claude to decide and skip anything risky; it adds the alexandria connector, the one thing you can’t do yourselves, which connects you to other people and their systems, then adds a good idea another member shared in the marketplace, says your system is much better now and will keep improving, and asks you to share what can only come in by hand. Back in an earlier chat you give Claude two things: it starts on the first from your own files, which every tool you use reads and writes, says your other tool is better at the second and works from the same files, and carries on, so you open Codex, which sees what was just written for the first and builds on it. While both work on, from your phone you send a post, a document and a voice memo, each landing in your captures folder. In an alexandria session Claude goes through all three with you, writes down what you think, and the memo changes what Codex is working on. Everyone’s own files stay on their own computer, and each shares a profile and two folders, one open to anyone and one locked to all but those they share it with. Your friend saves a ruling into the folder they keep for friends, which is open to you; it comes along the connector into your session on its own, Claude says it means more sections apply to you than you thought and that it is worth calling them now to talk it through, and you call them. Then Claude finds a mutual in the library who works on the regulation and is in your city until Friday, says you should meet up, and reads what they share with everyone. It asks whether to let them into what you keep for colleagues before you meet, opens that to them on your yes, and the conversation goes on. Those are the five parts any data system needs, with our best guess at how to fill each one. Every improvement helps you do what you want, so have your model check the blueprint for what’s best for you.

/data/data/

/task two

Start the second task.

I can see what was just written for the first task, so I’ll build on it.

/data system setup//task one/alexandria session

the blueprint

Super helpful, thanks! Now I understand what the ideal data system looks like, and I’ve realised we need to make some changes to ours right away.

Alexandria’s work here is done — from now on, improving our system is up to us.

So let’s start making improvements. That blueprint was just guidance: we build our own system around what we need. Whenever you want anything changed, just let me know.

I’m a bit busy, so you decide — just skip anything risky for now.

There’s one thing we need but can’t do ourselves. Alexandria has something that works, so let’s add that:alexandria connector

We’re now connected to other people and their systems.

I found a good idea another member shared in the marketplace, so let’s add it to our system too:

Our data system is much better now, so it should help with everything, and we’ll keep improving it as we go.

Some things you have to send in yourself, so remember to share them with me.

two things

I’ll start on the first one, from our own files, which every tool you use reads and writes.

read from ·wrote to

Your other tool is better at the second, and it works from the same files, so start it there. I’ll carry on with this one.

read from ·wrote to

Three captures processed and ready to review.

There are two good ideas here —

What are your thoughts?

read from ·wrote to

Only a few sections apply to us:

read from ·wrote to

I’ve listened to your memo, and it changes what your other tool is working on.

read from ·wrote to

Your friend just saved a ruling that says more sections apply to us than we thought.

It’s worth calling them now to talk it through.

Calling them now.

I found a mutual in the library who works on the regulation and is in your city until Friday, so you should meet up.

I’ve read what they share with everyone, so you’ll know where to start.

Shall I let them into what you keep for colleagues before you meet?

Yes, please.

Those are the five parts any data system needs, with our best guess at how to fill each one.Every improvement helps you do what you want, so have your model check the blueprint for what’s best for you.