How agent-ready is Gong?

Gong's website scored 40/100 on agentability, indicating moderate barriers for autonomous AI agents attempting to navigate, parse, and interact with its public pages. The product achieved consistent scores across homepage (40), pricing (43), and documentation (43), suggesting systemic gaps rather than isolated weak points.

The audit identified strong performance in machine readability (72) and control (70), meaning agents can reliably parse page structure and interact with UI elements. However, Gong received zero scores in both shadow UI avoidance and transparency, and only 17 in defaults. These weaknesses mean agents lack visibility into system reasoning, encounter UI elements that resist programmatic detection, and must navigate interfaces without helpful starting values or progressive disclosure.

For organizations deploying AI agents to research, compare, or integrate with revenue intelligence platforms, Gong's current implementation will require additional human oversight and custom parsing logic to work reliably.

Gong can improve agent-readiness by addressing the following transparency and defaults gaps:

• Expose an activity log or 'why this result' feature that records system actions in both human-readable and machine-readable (JSON) formats, giving agents visibility into decisioning and state changes.
• Add confidence scores or verified/best-guess labels to key outputs in API responses and UI, enabling agents to assess result reliability programmatically.
• Pre-fill form inputs with sensible defaults wherever reasonable values exist, reducing the decision burden on agents navigating configuration flows.
• Clearly mark which features and content are accessible without authentication versus requiring sign-in, so agents can plan interaction paths accordingly.
• Provide one-line summaries with expandable drill-downs for important results, and disclose detail progressively rather than rendering all fields simultaneously.
• Rephrase section headings as direct questions they answer (e.g., 'How do I cancel?'), improving both human scanning and agent intent-matching.