TL;DR
- The modern web was architected for human sensory perception — AI agents need something fundamentally different
- Anti-bot infrastructure, CAPTCHAs, and UI-patterns designed for humans actively break agentic workflows
- The shift toward an "agentic web" is not hypothetical — it is already underway, and businesses that adapt first will have a structural advantage
The Hook That Broke the Timeline
A few weeks ago, a single observation from Gregisenberg went viral: "Agents will be using the internet more than humans in 2-3 years."
It sounds dramatic. Then you look at the data and realize it might be conservative.
AI agents are already browsing product pages to compare prices. They're reading documentation to select libraries. They're scraping news to synthesize briefings. And that is just the beginning. As agentic systems become more capable and more autonomous, the volume of agent-to-web interactions will dwarf anything human traffic has ever produced.
The problem? The web was not built for them.
The Problem: Every Website Was Designed for Human Eyes
Walk through the average website with an agent's perspective and the cracks show immediately:
- CAPTCHAs are designed to gate content behind a test humans can pass and bots supposedly cannot. Agents can often solve them — but increasingly, they don't need to. Instead, they get blocked entirely, wasting compute and context.
- Cookie banners, consent modals, and infinite scroll are designed to manage human attention and regulatory compliance. For an agent, they are just noise that interrupts task execution.
- Anti-bot rate limits treat every programmatic request as hostile. Legitimate agent workflows — which may legitimately paginate, scrape, or query at scale — get throttled or banned.
- Dynamic content loaded via JavaScript requires agents to render pages the same way browsers do, dramatically increasing the cost of each interaction.
None of this was malicious. It was the right answer to a different question. The web was built for humans — and it has been optimized for human comfort, human attention, and human emotion for 35 years.
What's Actually Breaking
The friction is not theoretical. Teams building agentic systems today report a consistent set of problems:
Authentication and access. Agents frequently need to access content behind login walls. Session management, token refresh, MFA — these are all designed for humans who can read an SMS or click an email link. Automating them requires workarounds that are brittle and security-adjacent.
Structured vs. unstructured content. Humans can read a blog post and extract meaning from prose, layout, and context. Agents need machine-readable data. Most websites give them prose and hope they figure it out.
Pagination and navigation. Infinite scroll, "Load More" buttons, and multi-step form flows are intuitive for humans. For agents, they are obstacles that require page-state modeling just to traverse.
Rate limiting and blocking. When an agent needs to gather data from 10,000 product pages, it runs into anti-scraping defenses immediately. Legitimate agent traffic looks identical to malicious bots from the infrastructure's perspective.
The Infrastructure Gap: Who Builds Sites for Agents?
Here is the opportunity nobody is talking about loud enough: someone has to build the web infrastructure for agentic users.
Right now, there is a massive asymmetry. Businesses spend millions optimizing their sites for human conversion — SEO, UX, A/B testing, accessibility. Almost none of them are thinking about agent conversion.
The emerging answers are beginning to coalesce around a few themes:
MCP (Model Context Protocol) is one of the most significant developments in this space. It provides a standardized way for agents to interact with external tools and data sources — effectively an API layer purpose-built for agentic workflows. When a website exposes an MCP endpoint, agents can interact with it natively, without scraping or fighting through human-facing interfaces.
Agent-native APIs are starting to appear. Rather than serving HTML optimized for browser rendering, these APIs return structured data — JSON, Knowledge Graphs, or purpose-built schemas — that agents can consume directly. The site is not "a website" in the traditional sense. It is a data service with an agent interface.
Structured data and schema.org markup remain underutilized. Most sites that use structured data do so for search engine benefit. The same markup can serve agents equally well — if the site is designed with that intention.
Agent-aware rate limiting is an emerging concept. Rather than blocking all programmatic traffic, infrastructure that distinguishes between malicious bots and legitimate agentic users — based on behavior signatures, authentication, and intent signals — can offer different tiers of access.
What This Means for Businesses
The strategic implication is uncomfortable: your website is optimized for the wrong visitor.
You may have invested heavily in making your site intuitive for humans. That is not wrong — humans are still your customers, your users, your readers. But if the Gregisenberg prediction is even directionally correct, the composition of your audience is about to shift dramatically.
The businesses that win the next wave will ask different questions:
- Can an AI agent accomplish its task on our site without hitting a dead end?
- Is our product data available in a form agents can consume?
- Do we have an agent-facing interface, or just a human-facing one?
- When an agent is researching our category on behalf of a human decision-maker, does our content come out ahead?
This is not about replacing your human website with something else. It is about building an agentic layer — a parallel interface that makes your business accessible to the systems that are increasingly making decisions on behalf of humans.
The Opportunity: Build for the Agentic User Before Everyone Else Does
The agentic web is arriving faster than most people realize. The infrastructure is immature, the best practices are still being written, and the window for differentiation is open right now.
Every business that publishes content, sells products, or provides services online is effectively a content and commerce provider for both human and agent audiences. Most are optimized exclusively for the former. The ones that figure out how to serve the latter — without abandoning the human users they already have — will have a structural advantage that compounds over time.
The web was built for humans. It worked. Now it is time to build the next layer — for the agents that are becoming just as fundamental to how work gets done.
This piece was inspired by Gregisenberg's viral observation on the agentic web. If you want to think through what an agentic layer could look like for your business, explore Copanion — built for the era of high-agency AI.
