Issue 80Published October 25, 2026
Editor noteThis archived Ash AI Daily issue retains the delivered editorial briefing and final cards.
Story 1AI Daily
Block introduces Dessa, a speech-only conversation model
Block says Dessa is a full-duplex conversational model trained from scratch on human speech, with no text data or pretrained models. The company says it can listen and speak concurrently and run locally on a MacBook Pro M4 Max or better. This is a research preview: Block has not published an API, weights, formal benchmark suite, paper, or independent replication.
- Block says Dessa is a full-duplex conversational model trained from scratch on human speech, with no text data or pretrained models. The company says it can listen and speak concurrently and run locally on a MacBook Pro M4 Max or better. This is a research preview: Block has not published an API, weights, formal benchmark suite, paper, or independent replication.
Why it matters
Voice interfaces normally turn speech into text before generating a reply. A direct-audio approach could preserve timing, tone and interruption cues, but businesses should treat Dessa as early research until its methods, safety properties and performance are independently evaluated.
Sources
Block introduces Dessa, a speech-only conversation modelBlock — Dessa · Oct 25, 2026
