the last time I was in a meeting like this I was sitting in a conference room in Dongguan eating fresh lychees off a big bowl in the middle of the table in Brixton this time, and no lychees, but just as exciting

What if you could change device behaviour, just by talking to it? Create an app. Change it. Edit it. Add personal context. Then completely reinvent what the device is for. WORKS is our limited release of the first ever Promptable Products - physical devices inhabited by AI agents that understand what you want and do what you mean. WORKS launch in November 2026. This is a demo (using a ESP32 device from @M5Stack) showing one possible experience you can create. We’re building @inanimate_tech in public. Jump in, it's gonna be weird.

Join us in the @inanimate_tech lab as WORKS comes together. We shot this yesterday! WORKS is our limited release of Promptable Products at @inanimate_tech: physical devices inhabited by AI agents that understand what you want and change their behaviour when you ask. Right now... • Components arriving from around the world • Final design details taking shape • Provisioning and claiming process is nearly complete • Richer interfaces coming to our screens • Devices talking to each other and responding almost instantly This is WORKS in Progress: It's All Coming Together in the Lab

We let an agent express itself IRL through a device with a face. Weird. But we learned a lot. Meet Face P1, one of our more recent @inanimate_tech prototypes. Two screens for eyes, another for a mouth, a mic array, and a nose printed on the circuit board. @genmon asked it to scroll truisms across its mouth. It started “talking” to us, and those words became surprisingly interesting to reflect on. “Every system protects itself first.” Like: What do you mean? And why are you saying this now?! It’s teaching us about how people want to relate to agents and whether those are relationships we want our products to encourage, or whether we think they are problematic. This is Prototyping: "Why We Made an AI Device with Eyes and a Mouth."

Our first AI devices ship in weeks. Five WORKS named. Arc upgrades from Tetris to Mario. Global firmware updates now live. Plus: → @_lyraaaa_ Fruit fly brains playing Beat Saber → @typesafeai Jev’s sub-150ms decision models → @Musecases brings agents into hardware Lab Notes from @inanimate_tech, written by me + @gemon. Subscribe → news.inanimate.tech/p/lab-no…

We need to lock the designs for our five AI devices. Step inside the room where decisions are being made and biscuits are being eaten. WORKS is our limited release of Promptable Products at @inanimate_tech - physical devices inhabited by AI agents that understand what you want and change their behaviour when you ask. Shipping Autumn 2026. Here is me and @genmon working through the hard choices with our industrial designers. A screen we love but might struggle to source. A prototype that feels too big. Our logo looking good enough to temporarily distract us from the risks. We’re sharing the process as it happens, while there’s still plenty to figure out. This is WORKS in Progress EP5: The Hard Choices Behind Our AI Devices.

So @genmon is obsessed with air quality. While I'm not interested in CO₂ levels, I am very interested in what happens when you give an AI agent control of a fan. Is it a good idea? We built one to find out. At @inanimate_tech we learn by making things and living with them. Here's our latest prototyping experiment - number four in the series.

It’s OpenAI DevDay. Seems like a good moment to tell you what we’re building, too. 😘 Five AI devices. Made for you... and your agent. An excitable timer. A calendar that never forgets. A locked-in context bar. An agent console. And... an oracle. These are the archetypes - the five starting points - we’ve chosen for WORKS at @inanimate_tech. WORKS? Yep. Our limited release, fast-to-market AI devices - Promptable Products that do what you mean. Shipping later this year. Our bet: form matters. A clock gives you different ideas from a calendar. It also gives an agent different ways to act in the world. What you and your agent make of those possibilities is where things get interesting. Filmed on a stroll through the @BarbicanCentre. Slightly smaller production budget than DevDay. This is WORKS in Progress EP4: Decision Made - Five AI Devices We're Building

If an AI agent can turn a device into almost anything, how do you know what to ask for? A familiar shape gives you somewhere to start. At @inanimate_tech, we’ve been exploring this with Stick and Watch, off-the-shelf ESP32 dev kits from @M5Stack running our agent-generated code. A round screen suggests a clock, or a face. Two buttons on top suggest a stopwatch. Those familiar forms gave us ideas about what to ask our agent to make. @genmon made animated faces, an egg timer you could flip over, and I made a barbecue timer that became an unexpected breakthrough. The shape of the object helped us imagine what its behaviour could become. This is Prototyping: Why an “Anything” Device Needs a Familiar Shape.

New Lab Notes from @inanimate_tech: device simulators in-browser, launch planning underway, and @betaworks Demo Day video released. We’re bringing WORKS closer to production. Plus my take on Francis Bourgeois and enthusiasm as an antidote to cynicism. Subscribe ↓ news.inanimate.tech/p/lab-no…

Agent nooks, infinite calendars, birthday candles, vibeometers - and emoji machines. These are some of the device concepts we’ve been exploring for WORKS - our fast-to-market, limited-release promptable products, built in public and tested in the open. Each one offers a different way for agents to inhabit everyday objects. In the last episode, we took you inside the Approach design studio. This time, @genmon and I tour the sprawling Figma board behind the process - and the fifty concepts the team worked up. Throughout the design process, we’ve moved between methodical exploration and intuition. A framework helps you explore a direction. An odd reference sparks a conversation. Something catches your attention, and you follow it. Where did we end up? An agent with its own little nook on a shelf. A calendar that helps you see the flow of your day. An object that reflects the mood of a room back to you. We can’t build them all. We need to pick five. It's a balance between what people might want and what we can actually make. This is where we are with @inanimate_tech: sharing the possibilities while we’re still working out what they should become. This is WORKS in Progress EP3: Exploring Our Maze of AI Hardware Concepts.

This was our second major prototype at @inanimate_tech. It's a desk clock you talk to. When you talk to another person, you can tell whether you’ve caught their attention. Eye contact helps you know they’re listening. What might these cues look like in a physical device inhabited by an AI agent? Enter: TIME P1. This was our first prototype with a microphone. Our first device that you could prompt to change it's behaviour using just your voice. No keyboard or prototype web UI. Its small round screen let us experiment with that sense of attention, while its display and buttons gave the agent ways to respond to what we asked. Living with it also gave us room to discover what it could become. A request for a Tamagotchi ended up on the clock by accident, and the agent made it work using text and buttons. I started using it as a Pomodoro timer, hooked up to my task list on @BearNotesApp via MCP, counting completed tasks as I worked. These experiments are part of how we’re developing Promptable Products at Inanimate: exploring how agents translate intention into behaviour, and how these devices find a place in everyday life. This is Prototyping: "Why We Built a Desk Clock You Can Talk To."