How agent-ready is Greenhouse?

Greenhouse achieved an overall agentability score of 40/100 on its homepage, 41/100 on pricing, and 28/100 on documentation pages. This indicates that autonomous AI agents will encounter moderate to significant friction when attempting to parse content, navigate the site, and perform operations programmatically. The product's public web presence is currently in the early stages of agent-readiness.

The audit identified three principles as critical weaknesses: transparency scored 0/100, shadow UI avoidance scored 15/100, and defaults scored 33/100. These gaps mean agents cannot easily trace system decisions, may struggle with client-rendered or hidden interface elements, and lack helpful starting points for input fields. Moderate scores in chunking (41/100) and control (43/100) suggest that content organization and action safeguards need improvement. Stronger performance in machine readability (69/100) and status communication (65/100) provide a foundation to build upon.

Addressing transparency and control issues will have the highest impact on agent usability. Without audit logs, confidence indicators, or robust undo mechanisms, agents cannot reliably validate their actions or recover from errors, limiting autonomous operation in production environments.

Greenhouse can improve agentability by implementing the following changes:

• Expose an activity log or audit trail in machine-readable format (such as a JSON event log) with a visible interface that records system actions, enabling agents to trace decisions and verify outcomes.
• Surface confidence or certainty metadata on key outputs, such as adding a confidence field to API responses or displaying 'verified vs best-guess' labels in the user interface.
• Gate destructive actions behind explicit confirmation steps with distinct visual styling, and support undo or override functionality on completed actions to allow agents to recover from errors.
• Pre-fill form inputs with sensible defaults wherever reasonable values exist, reducing the decision burden on agents during data entry.
• Provide source references or input citations on generated outputs by attaching relevant links or identifiers that trace results back to their origin data.
• Fix heading hierarchy issues by using a single <h1> per page and ensuring contiguous heading levels without skips, improving structural parsing for agents.