Skip to main content
MilestoneResearch

We Published 91 Articles About Agent Readiness — Here Is What We Learned

91 articles. 500+ business scans. 50 verticals analyzed. The largest agent readiness dataset in existence — and five patterns that show up in every single vertical, from developer tools to dentists. Here is what the data says.

AH
AgentHermes Research
April 15, 202613 min read

The Numbers at 91 Articles

When we launched AgentHermes, the concept of “agent readiness” did not exist as a category. There was no framework for measuring how well a business could interact with AI agents, no benchmark data, and no content explaining what businesses should do. We built the framework, scanned the businesses, and wrote the content. Here is where we stand.

91
articles published
500+
businesses scanned
43
average score (out of 100)
50
verticals analyzed

The articles span vertical analyses (healthcare, construction, fintech, HR, restaurants, and 45 more), technical deep dives (error handling, caching, CORS, pagination, OAuth), protocol guides (MCP, A2A, x402, agent-card.json, llms.txt), platform comparisons (Shopify vs WooCommerce, Salesforce vs HubSpot), case studies (Stripe, Resend, Slack, GitHub), and executive guides (CTO, CMO). Each one anchors to specific dimensions and real scan data.

Lesson 1: Every Vertical Has the Same Five Problems

We expected each vertical to have unique challenges. Instead, we found the same five problems in every single one — from construction to fintech to dentists.

1. No API

78% of scanned businesses have zero API endpoints

The most common failure across every vertical. Restaurants, dentists, contractors, law firms, gyms — none expose structured endpoints. Their only "interface" is a phone number and a website designed for human eyeballs.

2. No Structured Pricing

91% lack machine-readable pricing

Pricing lives in PDFs, "contact us" forms, or sales calls. An AI agent trying to compare three plumbers cannot get a price from any of them without a phone call. D4 Pricing (0.05 weight) is the lowest-weighted dimension — because almost nobody has it.

3. Phone-Only Operations

64% of businesses have phone as only contact method

Booking appointments, getting quotes, checking availability, scheduling service — all require a phone call. AI agents cannot make phone calls. Every interaction dead-ends at "call us."

4. PDF-Heavy Content

47% serve critical business data as PDFs

Menus, catalogs, service lists, insurance forms, compliance documents — all locked in PDFs. AI agents can technically extract text from PDFs, but it is unreliable, slow, and loses all structure. A PDF menu with prices is worth nothing compared to a JSON endpoint.

5. No Structured Data (Schema.org)

69% lack any Schema.org markup

Schema.org markup (LocalBusiness, Product, Service, FAQ) is the minimum bar for AI discoverability. Most businesses do not have it. Their HTML pages have no semantic markup, no JSON-LD, no microdata. AI crawlers get raw text with zero context.

The implication: You do not need a vertical-specific solution. You need a horizontal platform that solves these five problems for any business type. That is exactly what AgentHermes builds: structured endpoints, machine-readable pricing, booking APIs, JSON data, and Schema.org markup — auto-generated for 15 verticals and counting.

Vertical Leaderboard: Who Scores Highest and Lowest

After scanning 500+ businesses, here is how verticals rank by average Agent Readiness Score. The spread is enormous: developer tools average 56 while construction averages 8. View the full interactive leaderboard.

Vertical
Avg Score
Top Scorer
Bottom
Developer Tools
56
Supabase (69)
Most open-source projects (20-30)
SaaS / B2B
44
Stripe (68)
Enterprise SaaS (25-35)
E-commerce
38
Shopify stores (45)
WooCommerce stores (30)
Fintech
35
Plaid (62)
Regional banks (10-15)
Healthcare
33
Telemedicine platforms (40)
Individual practices (5-10)
Real Estate
28
Zillow (45)
Individual agents (3-8)
Restaurants
22
Chains with APIs (35)
Independent restaurants (0-5)
Construction
8
Project management SaaS (30)
Individual contractors (0-3)

Five Lessons from 91 Articles and 500 Scans

Beyond the data, five strategic lessons emerged that shape everything we build next.

The 9 Dimensions Are Universal

We designed the 9-dimension framework for developer tools. It applies identically to dentists. D2 API Quality matters whether you are Stripe or a pizza shop — the question is whether structured interaction is possible. D7 Security matters whether you handle financial data or appointment bookings. The framework does not need vertical-specific dimensions. It needs vertical-specific weights, which is what our 27 scoring profiles provide.

Developer Tools Win Because They Build for Machines by Default

The top 20 Agent Readiness Scores are all developer tools: Supabase (69), Vercel (69), Stripe (68), Slack (68), Resend (75). They score high not because they optimized for agent readiness — but because they build APIs for programs to consume. JSON responses, structured errors, API keys, sandbox environments, OpenAPI specs. Everything an agent needs is a side effect of building for developers.

The Gap Between Knowing and Doing Is Massive

We wrote 91 articles explaining exactly what to fix. We built a free scanner that shows the exact score and specific improvements. We created the /connect wizard that generates an MCP server in 60 seconds. And yet: the businesses that read our articles still have not scanned themselves. Reading about agent readiness is not the same as doing something about it. The action gap is the real bottleneck.

AI Citation Requires Structure, Not Just Content

We publish these articles partly for SEO and partly for GEO (Generative Engine Optimization) — getting cited by ChatGPT, Claude, and Perplexity when people ask about agent readiness. What we learned: AI models cite pages with first-sentence answers, Schema.org FAQ markup, structured data tables, and definitive claims. Blog posts that meander get ignored. Every article we write now leads with the answer.

One Platform Will Not Fix Everything

We initially thought AgentHermes could solve agent readiness end-to-end: scan, fix, host. Reality: the businesses that need help most are the ones least likely to use a platform. The dentist who scores 5/100 is not reading blog posts about MCP servers. The real solution is embedding agent readiness into the platforms they already use — Shopify, Squarespace, WordPress, Toast, Calendly. Our adapters for Shopify and WooCommerce are the beginning of that strategy.

What Comes After 91

We are not stopping at 91. The next wave of content targets the verticals we have not covered in depth: government procurement, pharmaceutical supply chain, sports and entertainment ticketing, legal services, and maritime logistics. Each one will follow the same pattern — scan real businesses, extract real data, and publish the findings with actionable fixes.

But the bigger shift is from content to tooling. We wrote the guide. Now we need more businesses to actually scan themselves and act on the results. The free audit tool takes 60 seconds. The connect wizard generates an MCP server in under 5 minutes. The infrastructure exists. The gap is adoption.

If you have read this far, you know more about agent readiness than 99.9% of business operators. The question is whether you will do something about it. We wrote the guide. Now scan your business.

Frequently Asked Questions

Why 91 articles specifically?

We set a milestone target and hit it. The number is not magic — the coverage is. 91 articles cover every major vertical, every scoring dimension, every protocol (MCP, A2A, x402), executive guides, technical guides, platform comparisons, and case studies. It is the most comprehensive agent readiness resource that exists. We wrote it because nobody else was going to.

What is the single most impactful article for improving my score?

Start with "How to Improve Your Agent Readiness Score" (/blog/improve-agent-readiness-score) for the step-by-step playbook. If you are technical, "The Definitive Guide to Structured Error Responses" (/blog/structured-errors-guide) gives you copy-paste code that improves 20% of your score. If you are an executive, read the CTO or CMO guide for your role-specific actions.

How often do you update these articles?

As the data changes. We re-scan businesses regularly and update scores, statistics, and recommendations. When new protocols emerge (like x402 for micropayments), we publish new articles and update existing ones. The articles are living documents, not static blog posts.

Will you reach 100 articles?

We will pass 100 within weeks. But article count is a vanity metric. What matters is coverage: are there verticals, dimensions, or patterns we have not addressed? As of 91, the major gaps are government procurement, pharmaceutical supply chain, and sports/entertainment ticketing. Those are coming.


We wrote the guide. Now scan your business.

Get your free Agent Readiness Score in 60 seconds. See how you compare across all 9 dimensions — and get the specific fixes to improve.


Share this article: