How agent-ready is New Relic?

New Relic's agentability audit returned scores of 33/100 on the homepage, 42/100 on pricing, and 44/100 on documentation, indicating significant friction for autonomous AI agents attempting to parse and operate the site. The platform performs adequately on status feedback (65/100) and machine readability (54/100), but falls short in areas critical to agent interaction.

The most severe gaps appear in transparency (0/100), shadow UI avoidance (15/100), control (19/100), and defaults (23/100). These scores suggest that agents will struggle to understand output confidence, interact with UI elements reliably, safely execute or reverse actions, and work efficiently without manual input of common values. Improving these dimensions would materially increase New Relic's compatibility with agent-driven workflows.

To improve agentability, New Relic should prioritize the following fixes identified in the audit:

• Add confidence indicators to API responses and UI outputs, such as a numeric confidence field or 'verified vs best-guess' labels, so agents can assess reliability of returned data.
• Implement collapsible result summaries that provide a one-line overview with expandable drill-downs, and attach source references or input citations to generated outputs for traceability.
• Pre-fill form inputs with sensible defaults wherever reasonable values exist, reducing the need for agents to supply boilerplate configuration.
• Gate destructive actions behind explicit, distinctly-labelled confirmation steps rather than same-styled buttons, and support undo or override capabilities on completed destructive changes.
• Provide clearly labelled pause, cancel, or stop controls with accessible names like 'Cancel import' for long-running operations, avoiding icon-only buttons.
• Fix heading hierarchy by using a single <h1> per page and ensuring no levels are skipped, enabling agents to parse document structure correctly.