Short answer
Google CEO Sundar Pichai expects AI agents to reach websites in two ways: by browsing like a person and through protocols built for agents. He says both will grow, with protocols scaling better but taking longer to develop. Sites should prepare for both kinds of traffic.
Highlights
- Pichai expects agents to reach sites by browsing and through protocols, with both developing together.
- Agents running on Google's servers arrive as Google-Agent and generally do not follow robots.txt.
- Google's Universal Commerce Protocol lets merchants enable direct buying in AI Mode and Gemini.

2 min readAuthor: UNIT Journal EditorEditor-in-chief: Uğur Deniz İlhan
As Search Engine Journal reports, Google CEO Sundar Pichai described on the Sources podcast how AI agents will reach web content. An agent is software that browses, fills in forms or shops on a user's behalf, so how your site appears to agents is now an accessibility question.
How will agents reach websites?
Pichai sees two routes: agents that browse like a person and protocols developed specifically for agents. He called the browser route "very elegant" because it does what the user would do and keeps the user in the loop. Protocols, he said, are "a far more scalable approach" that could work anywhere and instantly, but will take longer. He noted "a lot of momentum" behind MCP, agent and payment protocols, while expecting a "chaotic evolution".
What tools does Google have here?
Google backs an open standard for agent shopping, the Universal Commerce Protocol (UCP); merchants can use it to enable actions such as direct buying in AI Mode and Gemini. On the browsing side, agents hosted on Google's servers arrive with the Google-Agent user agent. Google classes it as a user-triggered fetcher, and such fetchers generally do not follow robots.txt directives, so a robots.txt block will not usually stop them.
What happens to ad revenue when agents browse?
Asked how Google protects an ads business that needs engagement, Pichai said "the right answer longer term is you put users first" and that the market will find an equilibrium that will shift. His answer did not address how Google's ads would work when an agent does the browsing.
What should sites in Turkey do?
Agent traffic may be small today, but preparing is cheap. A set-up where product, price and stock information reads clearly to people and machines, and forms and the basket work simply, helps on both routes; reviewing your web and e-commerce infrastructure is a good start.
- Search your server logs for the Google-Agent user agent; Google publishes its IP ranges, but user-agent strings can be spoofed.
- Give price, stock and delivery information on product pages in structured data that matches the visible text.
- Keep basket and form steps simple: an agent browsing like a person passes through them like a person.
- If you use Google Merchant Center, follow options such as cart transfer with UCP; Pichai expects agent features to reach a wider audience over the coming months.
Frequently asked
- Can Google-Agent be blocked with robots.txt?
- According to Google, user-triggered fetchers generally do not follow robots.txt directives. A robots.txt block will therefore not usually stop these agents.
Sources
- Search Engine Journal ·
Follow UNIT Journal
What's new in search, AI and technology, in your feed every day.


