Pichai Says AI Agents Will Click Through Pages And Use Protocols
Mika Sandgrove | | 4 min read
Google is pointing to a web where software agents—not just people—navigate sites. Sundar Pichai said AI agents will “click through pages” and use “protocols,” while Google’s AI search surfaces keep changing how links appear without providing click data. Separately, reports suggest “AI visibility” may not be paying off (at least yet), as Google Ads expands Performance Max (PMax) creative sourcing.
What happened: Pichai says AI agents will click through pages and use protocols
Sundar Pichai said AI agents will click through web pages and use protocols, as reported by Search Engine Journal (10/22/2026).[1] Read literally, this is agent-mediated browsing: instead of a user clicking from a results page and doing the reading, an automated system could navigate page-to-page and complete information-gathering steps on the user’s behalf.
If that direction holds, “accessible” shifts from just indexable pages to reliable flows. Clean navigation, stable page states, and predictable pathways matter more when the consumer is an agent following a task.
Inline example: an agent lands on a category page, clicks into pricing, then extracts one constraint (a limit or fee) without a person hunting for it.
Timelines and implementation details weren’t included in the referenced snippet, so treat this as directional, not a rollout promise.
The measurement gap: AI Mode/AI Overviews links change again, click data still missing
Search Engine Roundtable reported (10/15/2026) that Google updated links in AI Mode and AI Overviews again—and still won’t show click data.[2] The issue isn’t only UI churn; it’s that teams still can’t validate referral volume from those AI surfaces with platform-native click reporting.
Practically, measurement shifts downstream: landing-page engagement, conversion rate, lead quality, and revenue signals. That helps with business accountability, but it doesn’t solve attribution to AI Mode/Overviews, which remains incomplete as long as click data is withheld.
Conflicting signals: studies suggest AI visibility/search may not be worth investing in (yet)
Search Engine Roundtable also flagged three studies (10/22/2026) suggesting AI visibility and AI search may not be worth investing in.[3] Without study details here (methods, definitions of “visibility,” segment splits), this is signal, not verdict—and outcomes likely vary by industry, query intent, and funnel stage.
What to watch
Budget-moving evidence looks repeatable: clearer platform reporting, consistent referral-to-conversion patterns over time, and controlled experiments that show incremental conversions (not just impressions or mentions). GEO talk is useful context, but it shouldn’t replace measurement discipline.[5]
Paid media update: Performance Max expands image and video sourcing options
Search Engine Land reported (10/22/2026) that Google Ads is adding new image and video sourcing options to Performance Max.[4] More automated sourcing can change what actually serves, raising brand-safety and compliance risk if asset governance is loose.
One practical implication to plan for: advertisers may be surprised by automation pulling from more sources.
Near-term action: review current PMax asset settings and recent top-serving creatives, then align approvals/brand rules to any expanded sourcing pathways once you confirm what’s live in your account.
Inline example: ensure brand-approved product images are eligible and complete so automation has “safe” inputs before it assembles variants.
What to do next (this week): monitor, don’t over-invest blind
Keep changes tied to what you can verify. Make core journeys agent-accessible—fast pages, stable navigation, fewer interaction traps that break flows. Assume AI Mode/Overviews attribution stays incomplete while click data is withheld, so stakeholder reporting should lean on on-site behavior and pipeline outcomes.
If you’re spending on “AI visibility,” keep tests small until you can show incremental conversions. In parallel, audit PMax creative sourcing and tighten governance so expanded automation doesn’t dilute brand or compliance.
Key Takeaways
Agents and AI surfaces appear to be moving faster than measurement. This week, the practical posture is readiness plus restraint: keep sites easy for software to traverse, avoid scaling investments you can’t attribute, and watch for two clarifiers from Google—more reporting on AI surfaces and more specificity on “protocols” for agent access—before shifting budgets materially.
Sources
- Search Engine Journal (10/22/2026) — Pichai says AI agents will click through pages and use protocols
- Search Engine Roundtable (10/15/2026) — Google updated links in AI Mode & AI Overviews again; won’t show click data
- Search Engine Roundtable (10/22/2026) — studies suggest AI visibility & AI search not worth investing in
- Search Engine Land (10/22/2026) — Performance Max expands image and video sourcing options
- Search Engine Land (10/20/2026) — GEO discussion context
Article author
Mika Sandgrove
Mika Sandgrove is an SEO writer and independent SEO consultant with more than three years of experience creating and optimizing content for search. He runs his own SEO practice, helping businesses improve their organic visibility through SEO strategy, content optimization, and technical and on-page SEO services. Much of his work comes through freelance marketplaces and online client platforms, where he works with businesses across different industries and markets. Mika primarily writes about SEO, search visibility, and practical optimization strategies, and is increasingly exploring Answer Engine Optimization (AEO) and how businesses can adapt their content for AI-powered search experiences.