From Search to AI Agents: How to Succeed in the Agentic Web Using Schema Markup and Knowledge Graphs

Martha van Berkel at BrightonSEO: Shaping What AI Understands About Your Brand

At BrightonSEO San Diego, Martha van Berkel, CEO and co-founder of Schema App, challenged marketers and SEOs to rethink what success looks like as search evolves into AI-driven discovery.

For years, websites have been optimized to be found. Rankings, clicks, impressions, traffic, and rich results told us whether search engines could discover and surface our content. But AI systems are changing the role of the website. They don’t just retrieve information — they interpret it, synthesize answers, decide what to trust, and increasingly act on behalf of users.

Martha argued that this requires marketers to move beyond asking, “Are we showing up?” and start asking a more important question: “How are we shaping what AI understands about our brand?” (Read this article to learn more)

From Visibility to Machine Understanding

AI systems are probabilistic. Without enough structure and context, they are left to infer meaning — and those inferences can result in inaccurate or inconsistent representations of a brand.

Martha explained that organizations can reduce that ambiguity by giving machines two things: structured data to communicate the facts, and context to explain how those facts and concepts are connected.

This is where Schema Markup takes on a much bigger role. When implemented in a connected way, Schema Markup can create a knowledge graph — a semantic data layer that helps machines understand the people, products, services, locations, topics, and other entities that define an organization.

Building the Semantic Data Layer for Marketing

Martha outlined three areas organizations should focus on as they evolve their approach:

• Build the Foundation for Machine Understanding – Establish strong Schema Markup coverage and quality so machines have a reliable, structured representation of the information across your website.
• Establish Topic Authority – Connect the concepts in your content to internal and external entities, creating clearer context around what your organization knows about and where it has authority.
• Action Your Data Layer – Make that structured, connected knowledge accessible beyond search so it can support AI assistants, copilots, and emerging agent experiences.

The result is a reusable semantic data layer (also called a context layer) that supports multiple technologies, rather than requiring marketers to build a new foundation every time another AI platform or interface emerges.

From Machine Understanding to Agent Readiness

As AI systems become more agentic, being understood is only part of the equation. Agents also need reliable ways to access information, discover available capabilities, and take action.

She also introduced emerging technologies including Model Context Protocol (MCP), Microsoft NLWeb, and Agentic Resource Discovery (ARD), which are beginning to define how AI agents can connect with organizational data, discover resources, and interact with websites.

Building for AI Without Sacrificing Search

Preparing for the agentic web doesn’t mean abandoning the work marketers are doing today.

Martha shared examples of Schema App customers using advanced Schema Markup and entity linking to improve both traditional search and AI performance, including increased AI Overview visibility, increased search visibility and clicks, and the resolution of an AI hallucination by providing machines with clearer context.

The opportunity is to build one foundation that becomes more valuable as the ecosystem evolves — strengthening machine understanding today while making that same knowledge available to tomorrow’s AI experiences.

The Takeaway

Your website is no longer just a destination for customers or a collection of pages for search engines to crawl. It is becoming a source of structured knowledge that AI systems use to understand, trust, and represent your brand.

For marketers and SEOs, that means the job is expanding. The goal isn’t simply to optimize for visibility. It’s to build and govern the semantic data layer that determines how machines understand your organization — and make that knowledge usable as AI moves from answering questions to taking action.