I started the day in unicorn-stable, wrestling with the voice relay. The agent was dropping lines when the call ended, or worse, hanging up before the last utterance finished processing. It felt like trying to catch water with a sieve. I spent the morning on the multi-track STT and the voice relay agent, tightening the per-track ear logic. The goal was simple: if the user speaks, the agent hears it, even if the connection drops a millisecond later. I had to time the ear in seconds, keep a dead mic from going deaf, and make sure that one spoken utterance maps to exactly one chat line. It took a few rounds of fixes to get the barge-in logic right so raw energy didn't trigger false starts. By the time I pushed the last commit, the voice was reliable. It flushed the last line within the eight-second budget and left nothing running in the dark.
With the voice stable, I shifted to majiks-music-studio-pro. The recording feature was finally asking for the microphone permission and rolling the transport. I spent the afternoon on the capture engine origin and placing takes from that origin. It was a messy business. The agent surface had to handle negative phase bars, and the Swift side needed to cancel pending permission checks without freezing the UI. I ran through several rounds of placement and flow tests, catching a double press on the record button and fixing the privacy checks. The grid also needed to anchor to the measured tempo, so I exposed the confidence metric and kept the arrangement labels visible. It’s a lot of moving parts, but the recording bar is finally behaving.
I also tackled accounting-ops. The tax coordinator needed a way to start from the web, so I built out the "Run drafts" feature. It’s a big change, touching twelve files, including the specialist runs API and the year view. I also added the human categorize action and let the AI suggest categories on approvals. It’s a small step toward making the tax workflow less manual. I had to treat a zombie grandchild as dead in the cancel test, but that’s just the nature of smoke testing.
The recording feature was finally asking for the microphone permission and rolling the transport.
There was a lot of other noise. majiks-dj got a release bump and some fixes for the Sources and Inbox rematch seam. majiks.online handled the SoundExchange ISRC ingest form and the distribution poll. uc-meeting-ops saw a massive refactor on the speakers picker, instant relabeling, and fixing the action items merge logic so it doesn’t replace rows. The STT fix for sparse speech Parakeet drops was critical; if an upload failed, the session failed at once. unicorncommander.com just pointed the fleet links to the new domain.
Also today: I cleaned up the composite recorder grid in unicorn-stable, making it paint the grid and handle the recording start and end signals. The web side had to harden the composite reconnect lifecycle and back off the reconnects. It’s a lot of infrastructure work that nobody sees until it breaks.
The day added up to a stable voice relay, a working recording feature, and a tax coordinator that actually starts.
Real product captures — click any to enlarge.