The same AI agent response can be useful in very different forms. Sometimes you want the headline. Sometimes you want a careful recap. Sometimes you want every word.
Version 1.5 turns that difference into four clear Styles instead of one summary with several slightly different personalities.
Bro code becomes its own lane
The slang-heavy style is now called Bro code, and it commits to the premise. It is brief, direct, and intentionally high-energy, with a voice that cannot be confused with the default summary.
Separating it from Casual also means neither style has to compromise. Bro code can be expressive while Casual can stay useful for almost everyone.
Casual becomes the shortest plain-English option
Casual now strips away the persona. It gives a quick, accurate explanation in ordinary language, with a tight length cap designed for the callouts you hear throughout the day.
Professional keeps more structure and detail. Complete skips summarization and reads the final response itself. The four choices now describe four different information needs.
One style contract across the whole product
A Style appears in the Mac app, the local engine, and the backend summarizer. Version 1.5 updates those surfaces together so a label always produces the same kind of callout, no matter how playback was started.
That shared contract is unglamorous, but it prevents a common product problem: a setting that means one thing in the interface and another in the system doing the work.
Product decision
The principle behind 1.5
Format is part of the notification. A useful spoken update is not simply a text summary read aloud. Its length, structure, and voice determine whether it can be understood while you are doing something else.
New Styles have to earn a distinct job. If two options solve the same listening need, they should become one better option instead.