How agent-ready is Make?

Make's agentability audit returned scores of 18/100 on the homepage, 14/100 on pricing, and 39/100 on documentation, indicating significant barriers for autonomous AI agents attempting to parse and operate the platform. The documentation page showed moderate structural readability, while the homepage and pricing pages presented substantial navigational challenges.

The audit identified critical weaknesses across multiple Agent Factors Engineering principles. Transparency, defaults, and shadow UI avoidance all scored 0/100, meaning the platform provides no machine-readable activity logs, lacks default values for agent interactions, and does not differentiate between visible and hidden UI patterns. Control scored 13/100, reflecting missing safeguards for destructive actions and limited interrupt capabilities. Machine readability reached 40/100, suggesting some structural markup exists but remains inconsistent across key pages.

These scores indicate that Make's current interface is built primarily for human users, with minimal affordances for programmatic or agent-driven interaction. Agents will struggle to confirm action outcomes, trace decision paths, or safely navigate critical workflows without substantial manual supervision.

Make can improve agent-readiness by implementing the following targeted fixes:

• Add semantic HTML landmarks (<header>, <nav>, <main>, <article>, <section>, <footer>) to replace generic <div> and <span> elements, improving structural navigation for agents.
• Expose an activity or audit log with both a visible feed and JSON event log, allowing agents to verify and trace system actions in machine-readable format.
• Gate destructive actions behind explicitly-labelled confirmation steps that are visually and semantically distinct from standard buttons, preventing accidental irreversible operations.
• Provide labelled pause, cancel, or stop controls for long-running tasks with accessible names like 'Cancel import' rather than ambiguous icons.
• Surface confidence scores or verification labels on key outputs through API response fields or UI indicators distinguishing verified data from best-guess results.
• Support undo or override functionality on completed destructive actions, with clear visual affordances for reversing recent changes.