AI Tone Assist: Catch Problematic Phrasing Before Feedback Is Submitted

Flo is proud to keep the updates coming, and this one addresses something that matters well beyond efficiency. AI Tone Assist is now available in Flo Performance Management, giving reviewers in-the-moment feedback on the tone and language of their review comments before anyone else sees them.

The challenge

Tone in a performance review is easy to misjudge. A comment can read as harsher or vaguer than the writer intended, and language that carries implicit bias can slip in without anyone catching it before the review is published. For Professional Development and HR teams, manually reviewing every comment for tone is not a realistic part of the process.

What we launched

Tone Assist analyzes reviewer comments in real time as they are written, highlighting specific phrases that may read as problematic and showing a flag card with an explanation and, where available, a suggested alternative. Reviewers can apply the suggestion to update the text or dismiss the flag to clear the highlight for that editing session.

The flags Tone Assist looks for fall into several categories:

• Gendered or coded language applied unevenly across groups (for example, "abrasive" or "too emotional")
• References to protected characteristics
• Vague or unsupported feedback with no specific example
• Unprofessional or inappropriate language
• Overused HR cliches such as "team player"

Flags are informational only. Tone Assist never blocks submission, and reviewers can always submit a form regardless of open flags. Dismissing a flag clears the highlight for that editing session but does not permanently resolve it.

For admins, access to Tone Assist is controlled by a role-level toggle in AI settings, with the same options as AI Rewrite: no one, admins only, or all users.

Why it matters

Performance reviews shape careers. When the language in a review is vague, coded, or inconsistent across groups, it can affect how an attorney is perceived, developed, and advanced at a firm. Tone Assist gives reviewers a chance to self-correct in the moment, before feedback reaches a reviewee or becomes part of a formal record.

For Professional Development and HR teams, this also provides a layer of protection that does not depend on manual oversight. Comments that touch on protected characteristics or apply coded language unevenly across groups are surfaced before a review is published, which reduces risk for the firm and supports a more consistent standard of feedback across the board.

Tone Assist is live for reviewers now. We think this one will generate real conversation among the teams using it, and we want to hear what you observe as it rolls out. Your feedback shapes where we take this next.

Want to see our latest updates in action? Book a demo and we'll walk you through them.