Not every discovery starts with a client. Some start in our R&D lab, with a question we couldn't stop thinking about:
How do you get live sound onto any screen in a room, reliably, without the whole show depending on one laptop or one piece of software?
This post is about work in progress. Nothing here has been used at a public event yet. But the lessons are already shaping how we build interactive work.
[ MEDIA SLOT 1 · VIDEO ] Short clip: sound playing, visuals reacting live in a browser. Replace this block with the video.
What we found
Audio-reactive visuals usually live inside one machine. A single computer listens to the sound, runs the visuals and drives the projector. That works for one screen. It gets fragile the moment you want several screens, phones in the audience, or visuals that keep running when something else crashes.
We also noticed something about our own process: AI coding assistants are excellent at writing JavaScript and shaders, but they can't build node networks inside visual tools like TouchDesigner. If AI is going to help us build faster, the core of the system has to be code.
What we built
A relay between sound and screens. Pure Data, an open-source audio tool, analyses the sound and sends simple values ("how much bass, mid, treble right now") over the network. A small relay service we built picks them up and sends them to any browser that is connected: a laptop, a projector, a wall of screens, a phone.
- It never floods the visuals. Only the latest value per channel is sent, at most 30 times a second.
- It runs on its own. It doesn't depend on our database or the rest of our platform, so the show keeps going if anything else goes down.
- It speaks both of Pure Data's message formats, so it works with existing patches.
A phone remote for TouchDesigner. For artists who do use TouchDesigner, we built a small app that turns the controls of a running project into sliders, toggles and buttons on a phone over the local network. Only the controls the artist chooses are exposed.
An interactivity experiment. A camera tracks your hand, your hand moves a 3D shape in the browser, and a local AI model shifts its mood. It's an early demo of where we want to take this.
[ MEDIA SLOT 2 · IMAGE ] Diagram or photo: audio in → relay → many screens. Replace this block, or delete it.
What we learned
1. Separate the listening from the showing. When sound analysis and visuals are decoupled, you can add screens, swap tools or restart a display without stopping the music.
2. Build the core in code you own. We chose a web-based stack we own over putting one commercial tool at the centre. AI can help write it, it runs on any screen with a browser, and every piece becomes reusable for the next project.
3. Respect the tools artists love. The phone remote exists because many artists won't (and shouldn't) leave TouchDesigner. Our job is to connect their work, not replace it.
4. Reliability is the feature. In a live show nobody sees the clever parts. They only notice when the screen freezes.
Who this is for
- Venues and bars that want visuals reacting to the music across several screens.
- Brands running activations that need interactive, audio-reactive moments.
- Creative studios and artists who need engineering behind their visual work.
- Permanent installations that must run unattended, night after night.
We can build this for you and your team
We deliver interactive systems as software, set up remotely, working with your venue or your crew on site:
- A system, not a one-off. Audio, sensors and visuals connected in a way you can reuse and extend.
- Your content, your tools. We connect what your artists already use.
- Kept running. For permanent installations, we can monitor and maintain the system after launch.
If you have a space, a sound and an idea, tell us what you want people to feel.