How agent-ready is Snowflake?

Snowflake's agentability audit returned scores of 39/100 on the homepage, 30/100 on pricing, and 44/100 on documentation. These results indicate that autonomous AI agents will encounter significant friction when attempting to parse information, navigate workflows, and execute tasks across Snowflake's public web presence.

The assessment measured performance across eight Agent Factors Engineering principles. Snowflake performed adequately in control (70/100) and status (65/100), showing that navigation patterns and system state are reasonably clear. However, three critical weaknesses emerged: transparency scored 0/100, indicating no machine-readable explanations for outputs or decisions; shadow UI avoidance scored 15/100, suggesting substantial reliance on client-side rendering that obscures content from agents; and defaults scored 18/100, meaning agents must supply values for fields that could reasonably be pre-populated.

These gaps create practical barriers for AI agents attempting to evaluate pricing, extract technical specifications, or automate workflows. Addressing the transparency, defaults, and shadow UI principles would yield the most significant improvements in agent-readiness.

The following improvements would directly address the most impactful gaps identified in the audit:

• Add JSON-LD structured data using schema.org vocabulary to describe primary entities on each page, enabling agents to parse product information, pricing tiers, and documentation topics without relying solely on visual layout.
• Implement transparency mechanisms such as confidence indicators on API responses, source citations for generated content, and an activity log that exposes system actions in both human-readable and JSON formats.
• Pre-fill form inputs with sensible defaults wherever reasonable values exist, reducing the number of decisions agents must make to complete common workflows.
• Clearly mark which features, documentation sections, and tools are accessible without authentication versus those requiring sign-in, allowing agents to route requests appropriately.
• Reframe section headings as questions they answer (for example, 'How do I cancel?' instead of 'Cancellation Policy') to improve agent comprehension of content purpose.
• Provide one-line summaries with expandable drill-downs for important results and decisions, giving agents the option to retrieve concise answers or detailed context as needed.