Drop a screenshot or paste a text exchange — saidwhat detects gaslighting, guilt-tripping, coercion, and other manipulation patterns using dimensional analysis, not keyword matching.
Built from analysis of 8,400+ labeled manipulation conversations across 6 types and 4 relational contexts.
Upload → Analysis → Conversation. After the analysis, you can ask follow-up questions. "Was there more context here?" or "What would a healthy version of this exchange look like?"
