AUTOMATIONSWITCH

Original Research

Research from Automation Switch

First-party datasets, scored audits, and the operator playbook behind every finding. Each wave fills in another vertical of the open-web map AI agents are learning to read.

Live

LiveWave 1 · Publishers

The Scaletific Agent-Readability Index, Wave 1: Publishers

103 publisher sites, 270 sampled articles

A scored audit of how legible 103 publisher sites are to AI agents. Five categories, ten signals, every one binary and detectable in code. The dataset, the rubric, the leaderboard, and the five fixes that lift any site forty points in a sprint.

Mean SARI score 49.9 of 100. Only 6 of 103 publishers have an llms.txt file. Only 8 of 75 with AI bot directives differentiate training crawlers from citation crawlers.

Upcoming waves

Each subsequent wave forks the Page Structure and Identity & Attribution categories of the SARI rubric for that vertical. The universal half (Discovery, Content Addressability, AI Bot Policy) ports unchanged.

PlannedWave 2 · E-commerce

SARI Wave 2: E-commerce

Same rubric, fork the Page Structure category for Product schema. 100 stores across Shopify Plus, regular Shopify, WooCommerce + Yoast, WooCommerce + RankMath, BigCommerce, native, Squarespace, and marketplaces. Cross-platform comparison: which stack ships agent-legible product pages by default?

PlannedWave 3 · Law

SARI Wave 3: Law and professional services

LegalService and Attorney schema. 75 firms across AmLaw 100, AmLaw 200, regional, boutique, solo, and legal-tech platforms. Practice area pages and attorney bio pages sampled. Where attorneys are cited by AI assistants and where they vanish.

PlannedWave 4 · SaaS

SARI Wave 4: SaaS marketing sites

SoftwareApplication schema. 100 sites across bootstrapped SaaS, VC-funded SaaS, enterprise SaaS, and developer tools. Features, pricing, and docs pages sampled. The agent-readability test for AI-assisted vendor selection.

PlannedWave 5 · Local services

SARI Wave 5: Local services

LocalBusiness and Service schema. 75 businesses across trades, hospitality, healthcare, professional services, and personal services. Where Apple Maps, Google Business, and AI assistants find a real local match, and where they make one up.

Methodology

Every research piece on this page uses the Scaletific Agent-Readability Index (SARI), a 100-point rubric across five categories. Three categories are universal across every site type. Two are forked per vertical. Every signal is binary or near-binary and detectable in code, so two auditors running the same script produce the same score.

  • Discovery (25 points) · llms.txt, sitemap, AI bot directives, MCP well-known
  • Page Structure (30 points) · Article / Product / LegalService / SoftwareApplication / LocalBusiness schema, per vertical
  • Identity & Attribution (20 points) · Author, brand, attorney, founder, or owner identity graph
  • Content Addressability (15 points) · Stable URLs, heading anchors, speakable / main block
  • AI Bot Policy Clarity (10 points) · Differentiated training-vs-citation directives