Most 360 and performance platforms now apply AI to written feedback, but they do different things with it.
Theme and sentiment analysis across comments. Culture Amp breaks feedback into sentiment, topics and trends. Qualtrics does the same through Text iQ, though its dedicated 360 product page mentions no AI at all. Workday Peakon surfaces themes and comment summaries, but documents this for engagement surveys rather than multi-rater reviews.
AI-written summaries of one person's 360. Effy AI generates a narrative summary per employee covering strengths, development areas and themes across all reviewers. Leapsome summarizes assessments with many contributors and suggests action plans. Betterworks summarizes a person's feedback over a set period. Lattice summarizes feedback and identifies trends inside the review draft.
Drafting rather than analysis. 15Five drafts the open-ended answers of a review from a year of work history. It does not claim sentiment or theme extraction.
No AI. SpiderGap's product pages mention none, and position report readability instead.
The more useful question is what the AI is allowed to decide. None of these vendors, Effy AI included, publishes an accuracy rate, an error rate or a hallucination rate for AI that reads employee feedback. Culture Amp is the only one to publish any figures, and they are relative to its own previous model — “25% more precise”, “30% fewer mistakes” — with no stated baseline or test set. On human control, 15Five is clearest: the AI cannot submit a review and the writer chooses every rating. Leapsome requires human review for rating adjustments. Qualtrics states the opposite for comment analysis, describing it as working “without requiring manual review”. Before you buy, ask who checks the AI's summary before the employee reads it.