Shannon

Why we are building Shannon

How much of your time do you spend on the things that are important to you? And is digital technology helping you spend your time on those things, or pulling you away from them?

If the research is to be trusted, the answer, for society as a whole, is that technology hinders us. Doomscrolling on feeds deliberately designed to consume as much of our time as possible, echo chambers that have reduced social cohesion, and unrealistic icons that wreck teenage self-image are three painful examples of distraction and even outright harm.

Now AI shows up at the door. The impacts on work are pretty clear: there will be more speed, more productivity. But the question every person should really be asking is whether it will make our personal lives better or worse.

As things stand, AI is far more likely to make it worse, especially in the hands of the same corporations that built our social media platforms. The same engagement logic that turned the feed into a money printing machine is being applied to AI, with significantly more powerful tools. We have already seen the first warnings. OpenAI released a version of GPT-4 so sycophantic that people couldn't stop talking to it, even as it was doing real harm to them. When OpenAI tried to fix that with GPT-5, the users rebelled, and they were forced to make the sycophantic version available again.

The research is starting to catch up to what most of us already feel. People who use AI heavily for companionship report more loneliness and weaker relationships with the actual humans in their lives, even as the conversations with the machine become more frequent.

So we asked: what would an AI look like if it was built, from the ground up, to serve you instead of the company that made it? An AI whose actual job was to push you back toward the people and things in your real life, not deeper into yet another feed?

That became Shannon's raison d'etre: Shannon is AI that helps you become the person you want to be, not the person an AI (or the company behind it) wants you to be.

In practice that means two things.

The first is just clearing the runway. Modern life carries an enormous amount of small admin: bills, inbox, calendar, reminders, school forms, appointments, the hundred tiny tasks that demand attention but give nothing back. None of it is the point of your life, but a great deal of life is consumed by it. Shannon will take as much of it as you let it.

The second is the more interesting half. Shannon thinks about your life in terms of seven broad areas: the people you are closest to, the values you hold, the things you love doing, the day-to-day running of your life, the community you belong to, your health and wellbeing, and the things you want to learn or grow into. As you talk to Shannon, it builds up a picture of you across these areas, and helps you turn the things you keep meaning to do into things you actually do. If you tell Shannon you want to read more, it works out that the half-hour after the kids are in bed is when you actually read, finds you the book it knows you would care about, and clears the bills and notifications that usually take that window away from you.

These seven areas aren't arbitrary. They reflect enduring ideas about human flourishing that appear across philosophical and spiritual traditions. We didn't invent these categories. We are reconnecting with a conversation that has been going on for a very long time.

Building it the right way

If we are going to ask people to let an AI into this much of their life, we have to hold ourselves to a higher standard than the platforms we are pushing back against. So here are the principles we are committing to, in plain language.

Your data is yours. We will not sell it, and we will not use it to train AI models without your explicit permission. At any time you can download everything Shannon knows about you, take it to another AI if you want, and delete the rest from us. If you decide to leave, we will make leaving easy.

We work for you, not for advertisers. Shannon is a paid subscription. That is by design. When the people using a product aren't the ones paying for it, the product gets built to serve whoever is. We would rather charge you a fair price than build the kind of business that has to harvest you to survive.

No dark patterns. Shannon is not engineered to be addictive. We do not optimize for time-on-app, and we do not measure our success by how many minutes you spend with it. If Shannon helps you live a better life with less of Shannon, that is exactly what success looks like to us.

A way out, not a way in. Shannon is built to push you back out into your life, not deeper into itself. It is not trying to become your best friend or your closest confidant. When it helps you turn back to the people right in front of you, it has done its job.

Energy is part of the bill. AI uses real electricity, and we think it's important to be honest about how much. By our current measurements, a year of typical Shannon use costs roughly the same amount of energy as a single three-mile school run. We will keep measuring, work to bring that down further, and give you the option to offset whatever's left.

Careful about the company we keep. Money has a moral weight. Where we can, we will avoid businesses whose work is built on harm: weapons manufacturers, mass surveillance contractors, or companies that profit from war and oppression.

These are commitments, not boasts. Some of them will be easy, others will be hard, and we expect to get a few of them wrong before we get them right. We would rather be measured against them than not be measured at all.

What we hope you do with it

Shannon is, in the end, a tool. The interesting question is not what Shannon does, but what you do with the time Shannon gives back to you. It might be the hour with your kid you didn't think you had this week, or the book that has been sitting on your bedside table for six months and finally gets finished. We don't know what it will be for you. The point of Shannon is that you do.

If Shannon turns out to be an example of an AI that serves you, instead of you serving the AI and the companies that make them, then we will have done what we set out to do.