🔥 SEOs Diners Club #216: Sundar Pichai: "Google Search Will Become an AI Agent Manager"

Hey everyone!

This week, Google CEO Sundar Pichai laid out his ten-year vision for Search on the Cheeky Pint podcast. What he said shook the industry: "Search will become an agent manager." Information-seeking queries will give way to task completion. Multiple agents will run simultaneously. This goes far beyond the "blue links are dying" debate.

In the same week, the March 2026 core update finished rolling out, and SISTRIX's Germany data painted a stark picture: 4 losers for every winner. Akamai's report documented a 300% surge in AI bot traffic, with publishers losing revenue directly. OpenAI announced a $100 ChatGPT Pro plan targeting Codex users. Google quietly released Eloquent, an offline AI dictation app on iOS. And Anthropic shipped Claude for Word, embedding AI directly into Microsoft's productivity suite.

Grab your coffee. This week demands strategic thinking.

🎯 Sundar Pichai's "Agent Manager" Vision

Search Is No Longer a List of Results

Sundar Pichai joined Stripe co-founder John Collison and investor Elad Gil on the Cheeky Pint podcast to discuss Google's AI history, infrastructure investments, and the future of Search. The episode was published on April 7. Marie Haynes summarized the most critical SEO takeaways in her detailed analysis on Search Engine Journal.

Pichai's most striking statement: "If I fast forward, a lot of what are just information-seeking queries will be agentic in Search. You'll be completing tasks. You'll have many threads running."

This is a far bigger claim than "blue links are dying." Pichai is saying that Search itself will become an orchestration layer. Users won't ask a question; they'll assign a task. Google will distribute that task across multiple AI agents and deliver the result.

Google Already Lives in This World: Antigravity

This isn't just a vision for the future. Google DeepMind and software engineering teams have already transformed their workflows using an internal AI agent tool known internally as "Jet Ski" and externally as "Antigravity." Pichai himself uses the tool, querying it with prompts like "We launched this thing. What did people think about this?" He said he used to spend far more time gathering information and that his life has gotten considerably easier.

2027: The Real Inflection Point

Pichai pointed to 2027 as a major inflection point when AI agents will operate with minimal human intervention, enabling fully autonomous workflows and reducing the need for manual prompting. He described delivering persistent, long-running tasks to users in a reliable and secure way as "the agentic future." He highlighted identity verification and access control as the fundamental building blocks of this future.

Search and Gemini Will Coexist

Pichai rejected the idea of replacing Search with a chatbot: "We are doing both Search and Gemini. They will overlap in certain ways. They will profoundly diverge in certain ways. I think it's good to have both and embrace it."

This reveals Google's internal strategy: Search and the AI assistant will remain two separate products, but they'll converge on task-based queries.

The $180 Billion Infrastructure Bet

Pichai confirmed the 2026 CapEx plan: $175-185 billion range. Google's CapEx has grown from $30 billion to approximately $180 billion. Pichai framed this not as speculative AI investment but as a response to observable demand: "We are supply-constrained. We are seeing the demand across all the surface areas."

Memory and wafer supply constraints mean that the full demand for AI capability cannot be met in 2026 and 2027. Even at $180 billion in annual CapEx.

Latency Budgets and the Gemini Flash Balance

Pichai described millisecond-level latency budgets operating at the sub-team level within Search. If a team shaves 3 milliseconds off an existing process, they earn 1.5 milliseconds of latency budget to spend on new features. Budgets range from 10 to 30 milliseconds depending on the type of work. Over the past five years, Search latency has improved by more than 35%.

Adding AI capabilities to Search without degrading speed remains an unsolved engineering problem. Gemini Flash models are managing this tradeoff: they operate at "90% the capability of the pro models" while being substantially faster.

Robotics and Drone Delivery

Pichai admitted Google entered robotics too early but said AI has become the missing ingredient for ideas conceived 10-15 years ago. Gemini Robotics models have reached state-of-the-art status for spatial reasoning. Wing drone delivery is scaling up, aiming to provide access to 40 million Americans within a reasonable time period.

Mert's Take: What Does "Agent Manager" Mean for SEO?

We need to take Pichai's vision seriously because this isn't just a prediction. It's a strategy backed by $180 billion in investment.

If Search truly becomes an "agent manager," SEO's fundamental question changes. It will no longer be "does our site rank when a user makes this query?" but rather "does the AI agent use our site as a primary source when completing this task?"

This aligns perfectly with last week's Kevin Indig research. AI citation isn't a writing quality problem; it's a content architecture problem. In agent-based Search, this becomes even more critical because agents won't just read your content. They'll use your data to complete tasks. Structured data, API access, and machine-readable content will be the SEO infrastructure of the next decade.

🔄 March 2026 Core Update: 4 Losers for Every Winner in Germany

Update Completed on April 8

Google confirmed that the March 2026 core update completed rolling out on April 8. The update started on March 27 and took 12 days. The December 2025 update had taken 18 days.

Google described it as "a regular update designed to better surface relevant, satisfying content for searchers from all types of sites." No companion blog post or specific goals were published.

SISTRIX Germany Data: 134 Losers, 32 Winners

SISTRIX's Germany analysis examined 1,371 domains showing significant visibility changes. After applying filters including 52-week Visibility Index history, 30 days of daily data, and visual confirmation, the picture emerged: 134 domains lost visibility, 32 gained. Roughly 4 losers for every winner.

Biggest losing categories:

E-commerce sites accounted for 39 of the 134 losers. Losses cut across verticals: fashion (cecil.de, down 30%), electronics (media-dealer.de, down 37%), gardening (123zimmerpflanzen.de, down 27%), and B2B supply retailers.

Recipe and food portals, already under pressure from Featured Snippets and AI Overviews, took further hits: kuechengoetter.de down 29%, schlemmer-atlas.de down 25%, eatsmarter.de down 18%.

Winners:

The biggest winner was audible.de at 172%, jumping from a Visibility Index of about 3 to over 8. ratiopharm.de gained 12%, commerzbank.de gained 11%, and government sites like hessen.de and arbeitsagentur.de gained 5-8%.

The clearest cluster signal SISTRIX identified was airport sites: Stuttgart up 22%, Cologne-Bonn up 18%, Hamburg up 17%, Munich up 8%. This may point to a broader ranking pattern rather than isolated site-level changes.

Information Gain as a Primary Ranking Signal

Information Gain scoring has become a primary ranking signal with this update. Google is now measuring how much genuinely new information your page contributes compared to content that already ranks for the same query.

Important nuance: the correlation between AI usage and ranking penalties is near zero (0.011). What gets penalized isn't using AI. It's using AI while abandoning quality control. AI-assisted content with human editorial oversight and original data remained unaffected.

🤖 Akamai Report: AI Bots Are Besieging Publishers

300% Surge, 96% Less Referral Traffic

Akamai's "Protecting Publishing: Navigating the AI Bot Era" report, published April 8, paints a disturbing picture. AI bot activity surged by 300% in 2025. The media sector attracts 13% of all AI bot traffic, and 40% of that targets publishers.

Here's the real problem: AI chatbots drive approximately 96% less referral traffic than traditional Google search. Bots scrape your content, generate their own responses, and answer users without sending them to your site.

Two Types of Bots, Two Different Threats

AI training crawlers: 63% of all AI bots targeting the media industry. They collect content to train large language models.

AI fetcher bots: Bots that pull content in real time to answer user queries. They represent 24% of AI bot activity targeting media, but publishers account for 43% of this segment. These bots pose a more immediate threat: they take your content and serve it to users in real time.

OpenAI Leads the Pack

A notable finding: OpenAI generates the highest volume of AI bot traffic targeting media companies. 40% of all OpenAI requests are directed at publishers.

Practical Steps

Identify AI bot signatures in your robots.txt: GPTBot, ClaudeBot, PerplexityBot, Google-Extended. Check your server logs to see which bots consume how much resources. Shift your content strategy from "drive traffic" to "become the primary source that AI cites."

🏥 Healthcare Reviews: Staying Compliant While Winning at Local SEO

Balancing HIPAA Compliance with Review Strategy

Search Engine Land's analysis tackles how healthcare businesses can manage patient reviews while maintaining HIPAA compliance and benefiting from local SEO.

Critical warning: merely acknowledging that a reviewer was a patient could constitute a HIPAA violation. Even if the patient revealed this in their review, the healthcare provider confirming it in a response is prohibited.

Key rules:

Focus on policy, not the person in your review responses. Address general facility policies and practices related to the complaint, not the reviewer's specific situation. Build a consistent, ongoing process: instead of one-off review campaigns, implement a systematic approach that collects at least one review per week. In the report's example, one facility reached nearly 500 reviews by February 2026, averaging at least one per week, without crossing any ethical boundaries.

💡 AI Developments

ChatGPT Pro: The New $100 Plan

OpenAI announced the $100 ChatGPT Pro plan on April 9. The plan targets Codex users and offers 5x more Codex usage than the Plus plan. Subscribers who join by May 31 get up to 10x increased usage.

The numbers are striking: over 3 million people use Codex weekly, a 5x increase over the past three months, with month-over-month growth exceeding 70%. The existing $200 tier continues (with 20x usage).

OpenAI positioned this plan as a direct competitor to Anthropic's $100 Claude option. The price and feature race in AI coding assistants is heating up.

Google Eloquent: Offline AI Dictation App

Google quietly released "AI Edge Eloquent," a dictation app, on the iOS App Store on April 6. No press release, no announcement. But the app has remarkable features.

Eloquent uses a Gemma-based on-device speech recognition model. It transcribes speech in real time without requiring an internet connection, automatically strips filler words, and transforms raw dictation into polished text. It includes four text transformation tools: "Key points" (bullet list), "Formal" (professional tone), "Short" (condensed), and "Long" (expanded).

Free, no subscription, no usage caps. Its competitors Wispr Flow and Willow charge $15/month. Users who sign in with a Google Account can import frequently used words from recently sent Gmail messages to build a personal vocabulary.

The Android version hasn't shipped yet. But the key takeaway: Google is making its on-device AI strategy concrete. Offline AI, combined with privacy advantages, is opening a new competitive arena.

Anthropic Brings Claude to Microsoft Word

Anthropic released the Claude for Word add-in in public beta. It runs as a native sidebar within Microsoft Word on Mac and Windows, available to Team and Enterprise subscribers.

Claude for Word enables direct drafting, editing, and revision within documents. Its most important feature: every change is surfaced as a tracked change you can accept or reject. Full control stays with you.

The add-in integrates with the previously released Claude for Excel and Claude for PowerPoint. A single conversation thread can span all three open documents simultaneously: compare data in a Word report against its Excel model, check consistency with PowerPoint slides, all within one AI session.

Anthropic is deliberately targeting the legal sector. Professionals working on complex documents like contracts, financial reports, and detailed memoranda are the primary audience. This move is part of a strategy to embed Claude directly into productivity tools rather than keeping it in a chat interface.

At Stradiji, we're starting to test Claude for Word for client reports and content revisions. I'll share our first impressions next week.

📊 6. This Week's Numbers

✅ Action Items for This Week

1. Start thinking from an "agent manager" perspective. Pichai's vision might seem distant, but Google AI Mode is already moving in this direction. Evaluate whether your content is not just readable but "usable" by AI agents: structured data, Schema markup, and llms.txt files are the first steps in this infrastructure.

2. Run your core update analysis (but don't rush). The update completed on April 8. Wait until around April 22 for Search Console data to settle. SISTRIX's Germany data can serve as a guide: e-commerce and recipe sites were the most affected categories.

3. Conduct an Information Gain audit. For each page, ask: "What does this content add compared to the pages currently ranking for the same query?" If the answer isn't original data, first-hand experience, or a unique perspective, the content is at risk.

4. Check your bot traffic. Search your server logs for AI bot signatures: GPTBot, ClaudeBot, PerplexityBot, Google-Extended. Update your robots.txt strategy accordingly.

5. Review your healthcare clients' review strategy. Build a systematic review collection process while maintaining HIPAA compliance. Never confirm a reviewer's identity as a patient in your responses.

👋 Closing

This week's core message: Search is evolving from "finding information" to "completing tasks."

Sundar Pichai's "agent manager" vision isn't just a prediction. It's a strategy backed by $180 billion in infrastructure investment. SISTRIX's Germany data shows the core update left 4 losers for every winner. Akamai's report documents AI bots scraping content while driving 96% less referral traffic than Google Search. Three different data sources point to the same conclusion: it's time to shift your content strategy from "ranking" to "becoming the primary source AI agents rely on."

Last week I summarized the formula as "sector-specific content architecture + comprehensive brand question coverage + multi-platform visibility." This week I'm adding to it: structured data + machine-readable content + Information Gain-focused original information = the 2026 formula for the agentic era.

If you want to adapt your SEO and GEO strategy to this new reality, we'd be happy to help at Stradiji.

See you next week. Take care!

Mert Erkal Stradiji | SEO, GEO & Conversion Optimization

Support the newsletter: if you find this content useful, buy me a coffee. ☕

About Mert Erkal Founder of Stradiji. 15+ years in SEO and GEO consulting for corporate clients globally. Author of SEOs Diners Club (English) and Dijital Pazarlama Notları (Turkish).

• Newsletter: seosdinersclub.beehiiv.com

• Twitter: twitter.com/merterkal

• LinkedIn: linkedin.com/in/merterkal

• YouTube: youtube.com/@stradiji