Blog
Read the latest news from the Chatmu blog.
Chatmu Desktop is an AI co-producer for Ableton Live: an agent that hears what Ableton is playing and acts on your whole set. You talk to it in plain language or with push-to-talk, it listens to your mix when you hit play, measures it, and then moves the faders, writes the MIDI and exports the stems. It launches today for Mac and Windows, starting with Ableton Live 11 and 12.
I'm Chema Rodriguez, founder of Chatmu. I'm a producer and sound engineer, I run a label with around 500 tracks, and I've spent years on the parts of a session nobody films: naming 40 tracks, gain staging, building buses, exporting stems for the mix engineer. Chatmu Desktop is the tool I wanted for that work, and for the nights when you need a second pair of ears.
Chatmu is an AI agent for the music industry with its own catalog of music-industry MCP connectors. Until now it handled what comes after the music: distribution, analytics, social, outreach. With Chatmu Desktop, it steps into the studio.
Let's answer the first question every musician asks: no, this is not Suno with a different logo.
Chatmu Desktop does not generate a finished song from a prompt. It works inside your Ableton session, on your tracks. When it composes, it writes MIDI clips you can open, edit or delete. When it mixes, it moves Ableton's own faders, sends and devices, the same ones you would touch. Nothing gets rendered into a black box.
That is the line we hold across the whole company: AI for the industry, humans for the music.
Setup takes a few minutes:
No API keys, no Python, no config files. It installs and works.
Hit play and Chatmu Ears listens. It measures loudness (LUFS and true peak), spectrum, dynamics and stereo image, and flags defects like sibilance, mud and masking. It runs null tests and loudness-matched A/B comparisons against a reference. Every number comes with a unit and a target, and the chat shows a spectrogram of what it heard.
Try: "Listen to the chorus and tell me what's wrong with the low end." The answer names the frequency range, the tracks fighting there and the fix it proposes, then waits for your yes before touching anything.
Create, name, arm and record tracks (with count-in); set faders, pan and sends; load plugins and change device parameters in real units; write MIDI clips and notes; manage scenes, locators and tempo; set up sidechain; move and group tracks.
Try: "Create an audio track called Guitar, arm it, and give me a one-bar count-in."
Chatmu writes parts in lo-fi, trap, reggaeton, house, R&B and pop, with drum grooves like dembow, drill, boom-bap, DnB and UKG. Everything lands as MIDI clips you can rewrite note by note.
Try: "Give me a dembow groove at 95 BPM and a bass line that follows the chords."
It mixes progressively, family by family (drums, melodic elements, vocals), routing each family into its own bus. You hear every step and approve it before the next one.
Export the full song, a time range, or up to 32 stems, saved straight into the artist's folder. Try: "Export the stems for the mix engineer."
Ask for a sound and Chatmu searches Splice for samples that fit the track.
The harmonic map reads your MIDI clips and gives you the exact chord in every bar, which is handy before you write a bass line or a topline.
Four plugins ship at launch:
| Plugin | What it is | What the agent does with it |
|---|---|---|
| Chatmu Coproducer | Bridge on the Master (VST3/AU) | Hears your mix and talks to the app |
| Chatmu Sound | Sound-design synth based on the open-source Vital engine | Programs sounds and recreates sounds from recordings |
| Chatmu Instrument | A playable synth the agent designs for you | Writes the sound and the plugin's face in real time |
| Chatmu FX | An effect the agent designs for you | Builds custom effects the same way |
The bridge. It sits on Ableton's Master and streams the mix to Chatmu Desktop, so the agent works from what it actually hears, not from guesses about your session.
A sound-design synth built on the open-source Vital engine. The agent programs it for you, and it can recreate a sound from a recording (a trumpet, a guitar, a bass) with pure synthesis and send it into Live as MIDI. Because it's synthesis, you can play it on any note and modulate it like any other patch.
Here the agent designs a playable synth for you in real time. It writes the sound from 31 DSP modules with 64 automatable parameters, and it even designs the plugin's face. The plugin hot-swaps without restarting Ableton. You get presets, 50-step undo and a library in your Music folder. Try: "Make me a dubstep bass with a pixel-art face."
The same idea for effects. Describe what you want and the agent builds it. Built-in examples include Tape Box and Plate Reverb.
If you've used Claude Code or Cowork, you know the pattern: the agent gets a working folder and keeps notes there, so every session starts with context instead of from zero. Chatmu Desktop does the same for music.
Each artist workspace has its own folder with a CHATMU.md file: notes about the artist and about each song, such as references, decisions you made and what's still pending. Exports are saved in that same folder. Open the session next week and Chatmu already knows where you left off.
If you've searched for ways to connect an AI assistant to Ableton, you've probably found some of these options. Each does something useful:
Chatmu Desktop puts the pieces together in one agent: it hears the mix, acts on the whole set, remembers each artist and song, and connects to everything else Chatmu does. If you already use Chatmu inside Claude or ChatGPT through the Chatmu MCP, Desktop is the studio side of the same account.
This is the part no plugin can do on its own. What you export from Ableton can go straight into the rest of Chatmu: a distribution draft with ISRC and cover checks, a lyric or visualizer video, and posts for your social calendar. Same agent, same artist workspace.
Try: "Export the final mix and start a distribution draft for Friday." Chatmu exports the file to the artist's folder and opens the release for review. It hears your mix and helps you release it. More on that flow in our guide to music distribution with MCP and in how to plan a release with AI.
Download Chatmu Desktop from chatmu.io.
No. It works on your session with you. It can write MIDI parts, design sounds and mix, but every decision is yours and everything stays editable in Ableton.
No. Suno generates finished audio from a prompt. Chatmu Desktop is AI inside your DAW: it writes editable MIDI and moves Ableton's own controls. It's not AI-generated music.
It starts with Ableton Live 11 and 12, on Mac (Apple silicon and Intel) and Windows. More DAWs are coming.
Yes. Exports are saved in your artist's folder on your computer. The stems and the music are yours.
No. Install the app, put the Coproducer plugin on your Master and start talking. It installs and works.
AI for the industry. Humans for the music.
