Why we built Ambi Today marks an important step for Ambi: our public launch. And today I also want to share why we built it. We are living through the shift from intelligence as a privilege to intelligence equality. Intelligence is no longer scarce. It should be used the way we use water, electricity, and gas — constantly, anytime, anywhere, in the most convenient way. But most people still treat it as something precious they only reach for when they “should.” They open a tool, explain the context, wait for an answer, and carry the result back to where life is actually taking place. Even the best demos still leave people on the street, at the table, in the kitchen — staring down at a glowing phone screen. That is a strange version of “anywhere.” I did not want another chat window with a smarter model. I wanted intelligence already in the room — like air. Continuously working, online and offline. Not trapped in a phone, a computer, or a Bluetooth bubble. And not something you batch-process later, after the day is over. That is why we built Ambi. What Ambi is Ambi is ambient intelligence that understands real-time context, builds memory, and helps turn intention into action. In an age of intelligence equality, the point is not to talk to AI more. It is to use intelligence to a greater degree — in every environment, in the moment, without stepping away from what you are doing. That means changing how people interact with AI: hardware and software working together, multi-device, and real-time — so intelligence comes into the environment, instead of living in asynchronous chat. A personal agent only works if it is there anytime, anywhere. That means two things at once: anytime-anywhere context, and anytime-anywhere ways in. Those ways in have to be fast and easy — or people will not use them. Real anytime, anywhere also means offline entry, offline context, and offline interaction — not just another chat surface on a lit display. We built Ambi around three ideas. Real-time context, so Ambi can understand what is happening around you across the devices you choose to use. Real-time interaction, so you can ask for help in the moment, without breaking focus on the present. And real-time action, so a thought can become a next step, a promise can become a follow-up, and the things that matter do not disappear into the rest of the day. Interaction alone is not enough. If intelligence is in the room with you, it should be able to act — not only chat. If you already need an always-on cloud entity that keeps state — a virtual computer for each person — you also need a running entity in the real world. Ambi is that bridge: it can turn any AI agent into ambient AI — giving it eyes, ears, and a mouth, so it can work continuously where life actually happens. Durable entry will not be another chat app with a different avatar color and a different name. An extension of yourself should not need its own personality or become a separate roommate. What matters is continuity — and the piece of hardware you actually wear. But bringing intelligence closer to our lives raises an important question: who remains in control? For us, the answer has always been clear. The person holds the direction. Intelligence carries the work. Ambi can prepare, organize, and suggest. You decide what happens next. You can accept, change, or dismiss what it proposes. The final decision remains yours. We are betting on real automation — proactive delivery where the human is not stuck in every loop — not on a better assistant that only answers when poked. And we are not chasing “more personality configs.” What matters is that Ambi is useful in your life — not that it performs a character. July 15 On July 15, we invited our first users to experience an early version of Ambi. More than 1,000 people signed up to take part. They used Ambi in real conversations and real workdays. They showed us where the experience felt natural, where it created friction, what they found useful, and what needed to be clearer. We listened. So we doubled down on real-time multiple agents — and grew that into Routines. While Ambi is listening in real time, multiple agents can work at the same time, completing agentic work for you as the moment unfolds — not after you go back to a chat window. Trust The closer intelligence comes to a person’s life, the greater the responsibility to earn their trust. That is why privacy is not a layer we add after building the product. It is part of how Ambi is designed — all the way through. Listening begins only when you choose to start it. You decide which services Ambi can access, and you can remove that access. Ambient intelligence should not keep raw audio. The same bar applies to every other kind of context — not just sound. We do not use your private content to train our models. We do not sell your data, and Ambi is not supported by advertising. Ambi should understand more about your life without taking more control over it. That principle will continue to guide us as the product grows. Today is an important step, but it is still only the beginning. We are opening Ambi gradually because we want to keep listening and learning as more people bring it into their lives. If you would like to be among the next to experience Ambi, join the waitlist. Anyone can join, and we will open access in stages as we learn, improve, and make sure the experience lives up to what we believe Ambi should be. And stay tuned! Our hardware will launch at CES in January next year. Ambi is intelligence for the life you are actually living — ambient, continuous, and already in the room. Use it like air. Peter