Imagine someone is looking for a tax advisor in Palma de Mallorca. In the past, they would open Google, type "tax advisor Palma," and click through ten blue links. Today, they tell ChatGPT: "Find me a tax advisor in Palma de Mallorca who specializes in startups."
The difference sounds small. But it fundamentally changes how your website gets found.
The Shift: From Searching to Delegating
What's happening right now isn't a theoretical future scenario. It's happening in real time.
People are increasingly delegating their research to AI assistants. Not everyone. Not for everything. But the trend is measurable:
- ChatGPT has over 200 million weekly users (as of early 2025)
- Perplexity processes millions of search queries daily
- Google itself integrates AI Overviews into search results
What this means for your website: There's a growing group of people who will never visit your website directly. Instead, they send an AI agent that searches, compares, and recommends on their behalf.
How Agent Traffic Works
An AI agent works differently from a Google crawler. Understand the difference, and you'll understand why AI-Ready matters.
Google crawler (since 1998):
- Reads your HTML code
- Indexes text, links, meta tags
- Evaluates based on 200+ ranking factors
- Shows your page in search results
- The human clicks and decides themselves
AI agent (since 2024):
- Receives a natural language query from the user
- Searches for websites matching the query
- Analyzes the page content
- If structured data exists: reads APIs and agents.json
- Summarizes, compares, recommends -- or acts directly
The crucial difference: An AI agent makes a pre-selection. The human doesn't get ten links but a recommendation. If your website doesn't appear in that pre-selection, you don't exist for that user.
What AI Agents Look for on Your Website
An AI agent isn't a human scrolling through your beautiful photo gallery. It looks for information it can process programmatically:
What agents understand:
- Structured data (JSON-LD, Schema.org)
- API endpoints with clear parameters
- agents.json with service descriptions
- Clear contact information in machine-readable format
- Prices, availability, opening hours
What agents cannot process:
- Text in images (infographics, flyer screenshots)
- Information hidden behind cookies or login walls
- PDFs without a text layer
- Flash or other proprietary formats (barely exist anymore)
What agents partially understand:
- Free text on your website (gets read, but structured data is more reliable)
- Reviews and testimonials (if marked as Schema.org)
- Prices mentioned only in body text (difficult to extract programmatically)
The Mobile Analogy: Why Timing Matters
In 2010, experts said: "Mobile is the future." Most companies nodded and did nothing. Their websites were desktop-only, and that was fine because most traffic came from desktop.
2013 was the turning point. Mobile traffic overtook desktop. Suddenly everyone needed a responsive website. Those who were prepared had an advantage. Those who weren't had to scramble -- often more expensive and worse than necessary.
In 2015, Google introduced the Mobile-First Index. Those without a mobile website were actively penalized.
The comparison with AI-Ready:
| Phase | Mobile | AI-Ready |
|---|---|---|
| "It's the future" | 2008-2012 | 2024-2025 |
| Early adopters benefit | 2012-2014 | 2025-2026 (we are here) |
| Turning point | 2013-2015 | Still open |
| Standard/mandatory | 2015+ | Still open |
We don't know when the AI-Ready turning point will come. Maybe 2027. Maybe 2029. Maybe sooner. What we do know: It will come. The direction is clear.
Honest Assessment: Where Are We Really?
No panic. No pressure. Here's the sober situation:
What's already real:
- Millions of people use AI assistants for everyday research
- ChatGPT, Claude, and Gemini can read and analyze websites
- Agent protocols like MCP (Model Context Protocol) are actively being developed
- Early standards like agents.json exist and work
What's not real yet:
- A dominant standard (there are several competing approaches)
- Measurable agent traffic for most small businesses
- Established analytics tools for agent visits
- A "Google for agents" that dominates the market
What this means: Those who become AI-Ready today don't do so because massive agent traffic is coming tomorrow. But because preparation is cheap now, the learning curve is gentle, and competition is still thin.
What Happens to Non-AI-Ready Websites
Here's a common misconception: Websites that aren't AI-Ready don't disappear. They'll continue to be found via Google, social media, and direct access.
But they become invisible to a growing audience. Specifically:
Scenario 1: User asks AI agent for a recommendation
- AI-Ready website: Agent finds structured data, understands the offering, recommends
- Non-AI-Ready website: Agent finds unstructured text, can't extract clear info, recommends someone else
Scenario 2: Agent compares providers
- AI-Ready website: Agent reads prices, services, availability via API
- Non-AI-Ready website: Agent tries to extract information from body text -- error-prone, incomplete
Scenario 3: Agent should act directly (book appointment, request quote)
- AI-Ready website: Agent uses API endpoint, completes the task
- Non-AI-Ready website: Agent can't do anything, tells user to visit the website manually -- friction that many won't bother with
The Numbers Behind the Shift
Reliable numbers on agent traffic are still hard to come by. That's a limitation we must honestly acknowledge. What we know:
- Gartner predicts: By 2028, 33% of interactions with digital services will run through AI agents
- ChatGPT usage: From 100 million to 200 million weekly users in less than a year
- Google AI Overviews: Already active for a significant share of search queries, reducing organic clicks
These numbers show a direction. They don't guarantee a pace. But they show the shift isn't hypothetical.
Practically: What Should Your Website Be Able to Do?
Don't worry -- AI-Ready isn't a complete rebuild. For most websites, it's manageable additions:
- agents.json -- A file that describes your services in machine-readable format
- Structured data -- Schema.org markup for your most important content
- At least one API endpoint -- e.g., for contact requests or appointment booking
- Health endpoint -- So agents know your website is reachable
- Clear, machine-readable contact data -- Name, address, phone, email in the right format
This isn't rocket science. These are technical standards that any capable web developer can implement in a few days.
The Strategic Question
The question isn't whether AI agents will be the next website visitors. They will. The question is when they'll account for a relevant share of your traffic.
If you prepare today, costs are low and time pressure is zero. If you wait until the turning point arrives, costs are higher and time pressure is real.
The mobile story showed: Those who prepared early don't automatically win. But those who come too late definitely lose.
Conclusion
AI agents are becoming website visitors. Not tomorrow and not all at once. But the trend is as clear as mobile was in 2010.
The smart response isn't panic but preparation. Implement structured data. Provide a few API endpoints. Create an agents.json. All doable, all affordable, all now -- before the market demands it.
Your website doesn't just need to be beautiful. It also needs to be machine-readable. That's always been the case (SEO is nothing else). But the machines doing the reading are getting significantly smarter -- and significantly more demanding.
