The Agentability Blog
Field notes and data from auditing how agent-ready the web is — what AI agents can and cannot operate, and what to fix.
WordPress.com's Agentability Challenge: What the Data Reveals About AI Agent Access
WordPress.com scores 43/100 for agentability, placing it in the Developing tier. Like most SaaS platforms, it faces critical challenges in transparency and shadow-UI avoidance that limit AI agent effectiveness.
Freshworks and the Agentability Gap: What 100 SaaS Homepages Reveal About AI-Ready Design
An audit of the top 100 SaaS products shows most aren't ready for AI agents. Freshworks and its peers face a common challenge: making software navigable by machines, not just humans.
The clean handoffs gap: what most top SaaS products get wrong for AI agents
Clean handoffs—how well software returns control to AI agents—averages just 40/100 across the top 100 SaaS homepages. One in five products scores below 20, creating a critical barrier to agent-driven workflows.
The sensible defaults gap: what most top SaaS products get wrong for AI agents
Across the top 100 SaaS homepages, sensible defaults averages just 35 out of 100, with nearly a quarter of products scoring under 20. This gap represents a critical barrier to AI agent operability that product teams can systematically address.
The shadow-UI avoidance gap: what most top SaaS products get wrong for AI agents
80 of the top 100 SaaS products score under 20/100 on shadow-UI avoidance, creating a critical barrier for AI agent interaction. Product teams need concrete strategies to close this gap.
The transparency gap: what most top SaaS products get wrong for AI agents
An analysis of 100 leading SaaS homepages reveals that transparency—the principle enabling AI agents to understand what actions are possible—scores just 5 out of 100 on average. This gap represents a fundamental oversight in how products communicate their capabilities to autonomous systems.
Machine-readable, but not agent-ready: what 100 SaaS audits reveal
The top 100 SaaS products have mastered machine readability but score near zero on the agent-era principles that enable verification and action. A new audit reveals the invisible gap between readable interfaces and truly agentic software.