MillerKnoll AI
Architecture · interactive

How MK Comment Signal is built.

Turns thousands of open-text engagement-survey comments into ranked themes with real verbatim evidence — PII removed, ready for a leadership deck.

MK Comment Signal — comments to themes

Model
GPT-5 Chat
Source
Perceptyx
Live
Jan 20, 2026
Users
HR / People

Upload the Perceptyx survey-comment PDF; a topic flow extracts the text for four survey questions; a series of analysis steps identify 4–5 recurring themes per question ranked by comment frequency, each backed by at least 12 verbatim quotes — with names, emails, and IDs auto-redacted. An optional Power Automate flow exports a formatted Word document to OneDrive.

Analyze this quarter’s engagement-survey comments
SourceTool

Hover or tap an agent above to see its pipeline and what it returns for this bid.

◆MK Comment Signal— Theme extraction + evidence

Transforms open-text engagement comments into structured, defensible insight. It extracts text from the Perceptyx export, identifies the recurring themes per question, ranks them by how many comments support each, and attaches verbatim evidence — while automatically redacting names, emails, and employee IDs so nothing personal leaves the analysis.

  • SourcePerceptyx survey-comment PDF · uploadedopen-text responses, 4 questions
  • ToolTheme identification · per question4–5 themes, frequency-ranked
  • ToolPII redaction · alwaysnames · emails · employee IDs removed
  • ToolWord export to OneDrive · optionalvia Power Automate
Input →
UploadPerceptyx comment export (PDF)
Output ←
Themes4–5 per question, ranked by support count
Evidence≥ 12 verbatim quotes per theme
SafePII redacted; optional Word doc to OneDrive

Why it matters · What a person would spend days reading and coding becomes a ranked, evidence-backed theme set — with the privacy handling built in.