Portfolio · Selected work

Technology and artificial intelligence, turned into product.

We're a technology development lab. We build platforms, software, AI agents, instruments and experiences — from idea to finished product — for global brands and for ourselves.

8Years
14Projects
4Platforms
4Global brands
scroll
What we build AI agents Back-end & APIs Web experiences iOS & Android apps Virtual instruments Audio plugins · VST · AU AI models Desktop software

We take the idea and turn it into a tool. Agents that talk, software that processes, platforms that create.

We work side by side with brands that want to turn an idea into an experience, and we build our own products when we spot a problem that hasn't been solved yet. Here are the most relevant launched works from recent years.

01Software development · Desktop · iOS · Android

Coca-Cola

Coke SoundZ — the instrument generated from "Coca-Cola sounds" using AI.

Coca-Cola took the sounds of Coke Zero Sugar — the "pssht", the fizz, the ice, the sip — and turned them into an AI musical instrument. We developed the software behind it: the desktop version (VST · VST3 · AU) and the iOS and Android apps, with AI sound generation, pads, mixing sliders, presets and a connection to the physical version, shaped like the bottle.

iOS AppAndroidVST · AUSamplesHardware sync
View project →
coca-cola.com/offerings/cokesoundz
Coke SoundZ
Coke SoundZ devices
02Virtual instrument · Desktop · iOS · Android

Kia

Kia Instrument — a synth made from the sounds of nature in motion.

For the "Sounds in Nature" campaign, Kia recorded the sounds of nature in motion around the world — thunder, geyser, forest, wind, water — and used them to create an instrument designed around the flow state. We developed the instrument's back-end and front-end: a sample-based synthesis engine, mixer, filter, ADSR, VCO, waveforms and keyboard, for desktop, iOS and Android.

Synth engineiOS · AndroidDesktopDSP
View project →
kia.com · sounds in nature
Kia Instrument keyboard
Forest — nature sound for the Kia instrument
03Full back-end · AI agent · Web

Johnson & Johnson

Speak to B-Cell — a conversation with the B-cell collective.

For Johnson & Johnson, we developed the full back-end of an experience where the user talks, in natural language, with the "collective consciousness" of B-cells. An AI agent grounded in immunology knowledge, inside an immersive interface. We're responsible for the entire back-end. The project is still in development.

Full back-endAI agentConversationalWeb
In development
speak to b-cell · johnson & johnson
Speak to B-Cell — Johnson & Johnson
04Product · AI for culture

Tarsila.ai

Artificial intelligence for creating cultural projects.

An AI that helps get cultural projects off the ground — from concept to structure. Built for cultural producers in Brazil, it turns ideas into well-formed projects, cutting the busywork and keeping the focus where it matters: creation. We're responsible for the entire back-end.

Applied AICultureText generationWeb
Visit site →
tarsila.ai
Tarsila.ai
05Own product · AI platform

CubeSoundLab

Describe an effect in text and get a ready-made plugin in seconds.

A pioneering ecosystem for creating audio effects with AI. You describe an audio effect in natural language — "an analog delay with repeats that degrade" — and get a working plugin in VST, VST3 or AU, without writing a single line of code. The platform also does AI music, stem separation and voice removal. Full project development.

Generative AIVST · VST3 · AUStem separationWin · MacNo-code
Visit site →
cubesoundlab.com
CubeSoundLab
CubeSoundLab effect chain
06Own product · DAW in the browser

LetsDAW

A complete multitrack studio that runs right in the browser.

Recording, mixing and mastering with nothing to install. Up to 16 tracks in real time: load audio, record, mix and master using VSM plugins, import and export sessions and render the final track — all in the browser, Linux included. Built-in LUFS, true peak and RMS metering. Full project development.

Web AudioVSM pluginsLUFS · RMSZero install
Open LetsDAW →
audiofb.com/letsdaw
01
02
03
04
05
MASTER-3.2 LUFS
07Plugin · Built-in AI model

RoomDIY

Acoustic spaces generated from room geometry.

A plugin that generates acoustic environments dynamically — no pre-recorded impulse responses. You draw the room, set materials and place up to 8 sources; it computes reflections and attenuation in real time. A neural network runs locally and generates acoustic reports: RT60, room modes, frequency response and treatment recommendations. Full project development.

Real-time acousticsLocal neural networkRT60 · modesVST3 · AU · AAX
View project →
audiofb.com/plugins/roomdiy
RoomDIY room editor
RoomDIY acoustic analysis
08Sound installation + site · Kyoto, Japan

7days / 7years

Japan seen through two lenses of time.

An exhibition at KYOTOGRAPHIE 2026, in Kyoto: Iris Fuzaro photographs Japan in 7 days, on film; Erick Rosa documents 7 years of routine, on digital. Two timescales in dialogue — the fleeting and the enduring. We developed the project's website.

Spatial soundWebAudiovisual integration
Visit site →
7days7years.com
7days/7years exhibition in Kyoto

Got an idea for your project? Let's build it.

From idea to finished product — AI agent, platform, app, instrument or experience. Tell us what you have in mind.

[email protected] audiofb.com