`Voker`: analytics layer for AI agent operations

Tracks agent outcomes through Intent, Correction, and Resolution, not raw logs. The 90%+ complaint-driven debugging stat signals a real gap between observability and product analytics.

• —The pitch is not trace debugging; it turns conversations into outcome metrics teams can act on.
• —A reported 90%+ of YC respondents learn failures from customer complaints first. That is a painful enough workflow to buy against.
• —Intent / Correction / Resolution is a useful product schema for any agent feature, even if you do not buy this tool.
• —The substitute today is dumping logs into Claude or ChatGPT for summaries. A dedicated analytics layer wins when trend detection and counting matter.