Google CEO Sundar Pichai has outlined a practical view of how AI agents may interact with the web: some will use websites much like a human visitor, while others may connect through protocols that link directly to services and APIs. Although this is not a formal product announcement or a new standard, the comments are relevant for businesses planning for AI-driven search, automation and digital customer journeys.

What happened

According to coverage by Search Engine Journal, Pichai said AI agents are expected to access websites in two ways: by browsing pages and by using protocol-based connections. He also indicated that protocol-based access may scale better over time, but could take longer to develop.

This distinction matters because it suggests AI agents may not rely only on traditional web pages. In some cases, they may read site content, navigate interfaces and interpret pages similarly to a user. In others, they may interact more directly with structured systems, such as APIs or service endpoints, if the ecosystem develops the required protocols.

At this stage, these remarks should be understood as an indication of direction rather than a confirmed rollout timeline, technical standard or binding requirement for publishers and businesses.

Why it matters for European businesses

For many European companies, websites currently serve two core functions: they provide information for human visitors and they support visibility in search engines. AI agents could expand that role. In future customer journeys, agents may compare products, retrieve business information, complete support tasks or initiate transactions on behalf of users.

If that happens, businesses may need to think about two layers of access:

• Page-level access: whether site content is readable, structured and usable when an AI system navigates the website directly.
• System-level access: whether product data, booking functions, support workflows or other business services can be accessed through APIs or similar machine-readable connections.

This has implications for SEO, e-commerce, content design, technical architecture and automation. Companies that depend on web traffic may need to prepare for scenarios where AI systems consume information without behaving exactly like a traditional search engine crawler or a human visitor.

For service businesses and software providers, protocol-based access could eventually matter even more. If AI agents can interact directly with booking systems, knowledge bases, product catalogues or customer service tools, the value of well-structured back-end access may increase.

Who may be affected

• E-commerce companies that rely on product pages, pricing information and on-site discovery.
• Marketing teams responsible for organic visibility, content performance and search strategy.
• IT and web development teams managing APIs, structured data, site architecture and platform integrations.
• SMEs adopting automation that may want AI assistants to interact with internal or customer-facing systems.
• Digital agencies and SaaS providers building websites, commerce systems or connected services for clients.

European businesses in regulated sectors may also need to consider how any future agent access model intersects with security, authentication, privacy and compliance obligations, especially where customer data or transactional systems are involved.

What companies should consider

• Review website usability for machine-driven navigation: Clear structure, accessible navigation and consistent page elements may become more important if AI agents browse sites directly.
• Strengthen structured content and data: Product details, service information, policies and business data should be accurate and well organised so they can be interpreted more reliably by automated systems.
• Assess API readiness: Businesses with bookings, inventory, account functions or service workflows should evaluate whether their systems can support secure, controlled machine-to-machine access if industry protocols mature.
• Plan for authentication and permissions: Direct agent access to services raises practical questions around identity, authorisation, rate limits and abuse prevention.
• Monitor AI search developments: Marketing and digital teams should watch how Google and other platforms handle agent-based discovery, referral patterns and content usage.
• Coordinate SEO, development and security teams: The shift is not only about content visibility. It may also affect infrastructure, governance and customer experience.

For now, most businesses do not need to rebuild their digital stack based on a single executive comment. But the direction is clear enough to justify internal planning. Companies that already invest in structured content, robust APIs and secure digital workflows are likely to be better positioned if AI agents become a more common way for customers to discover information and complete tasks online.