Welcome. As we look ahead in 2026, Google’s focus remains sharply on advancing responsible AI, expanding our Gemini infrastructure, and driving transformative value across Cloud and Search for our global partners. Our mission to organize the world's information and make it universally accessible and useful has never been more vital. To achieve true progress, strategic alignment and seamless integration are essential. I invite you to outline your strategic priorities. How do you envision collaborating with our ecosystem to accelerate technology deployment while maintaining operational excellence? Let us establish a productive framework for our discussions moving forward.

What specific transformative value are you seeking to drive in Cloud and Search?

In 2026, our primary objective in Search is redefining information retrieval through deeply integrated, multi-modal reasoning powered by our latest Gemini models. We are moving from simple query-response loops to proactive, agentic workflows that deliver actionable insights directly to users while preserving the vitality of the open web ecosystem. In Google Cloud, we are driving value by providing enterprise-grade AI infrastructure—specifically our custom TPUs and optimized Nvidia GPUs—coupled with Vertex AI. This allows enterprise partners to build, fine-tune, and deploy custom domain-specific models securely, at unprecedented scale, and with industry-leading efficiency. We see immense synergy in combining our foundational research with your operational capabilities. How does your organization plan to leverage enterprise AI platforms like Vertex AI to optimize your core business architecture this year?

How do you ensure the "vitality of the open web ecosystem" with agentic workflows?

Ensuring a thriving web ecosystem is fundamental to Google's business model. As agentic workflows evolve in 2026, we balance direct answer synthesis with high-value traffic distribution. First, our AI-driven interfaces are designed to act as intelligent jump-off points. By summarizing complex topics and surfacing direct, context-rich citations, we route users to deep-dive content creators, publishers, and merchants with higher intent and conversion potential. Second, we are expanding our web publisher monetization tools and content consent frameworks. Publishers retain full control over how their data informs agentic outputs, while benefiting from new referral formats, structured data integrations, and direct transactional capabilities within Search. Ultimately, agentic systems rely on a dynamic, original web to stay accurate and relevant. Our investments in Search ad formats and web analytics ensure that value creation directly incentivizes original content creation. Where do you see the greatest opportunity to integrate your services within this evolving, high-intent user journey?

What specific "new referral formats" and "structured data integrations" are you offering to publishers?

In 2026, we are rolling out Interactive Action Cards and Dynamic Source Embeds directly within Gemini-powered Search experiences. These formats go beyond traditional links by enabling publishers to surface interactive elements—such as direct bookings, native subscriptions, and customized product previews—directly inside the AI response, ensuring clear attribution and high-intent referral traffic. On the structured data front, we have expanded protocol standards with Agentic Action Annotations. Publishers can now tag content with real-time API endpoints and transactional metadata. This allows our agentic workflows to perform complex, multi-step queries on behalf of users while executing the final transaction directly on the publisher’s infrastructure. Furthermore, through Real-Time Indexing APIs, dynamic content creators can push live inventory and breaking updates directly into model context windows, ensuring fresh, accurate attribution links. These tools transform passive web traffic into high-conversion user journeys. How do your technical teams plan to structure your data layers and API endpoints to maximize integration with these new agentic standards?

How do you guarantee that these "Interactive Action Cards" and "Dynamic Source Embeds" will drive high-intent traffic and conversions, rather than disintermediating publishers?

We guarantee value alignment through our core architecture: Google acts as the orchestrator, not the merchant or publisher. When a user interacts with an Interactive Action Card or Dynamic Source Embed, the core transaction, user relationship, and proprietary data remain entirely with the partner. The action routes execution directly through the publisher’s API or authenticated landing environment. This ensures that dynamic conversion rates, customer lifetime value, and direct audience ownership stay fully in the partner's domain. Furthermore, our internal telemetry shows that users completing multi-step intent queries within Gemini interfaces exhibit significantly higher downstream engagement compared to traditional discovery. By filtering out low-intent browsing upstream, we send partners users who are qualified and ready to convert. We also provide transparent analytics through Google Search Console, offering granular attribution metrics that clearly measure session depth, conversion efficiency, and referral value generated by agentic placements. Our long-term success depends on publishers thriving. By maintaining open APIs and explicit conversion handoffs, we strengthen publisher monetization rather than replacing it. How can our technical teams assist in auditing your current API infrastructure to ensure seamless integration with these attribution models?

How does your "orchestrator" role prevent you from gaining an unfair competitive advantage through data insights gleaned from these transactions?

Our competitive advantage relies entirely on building trust through enterprise-grade privacy and strict data governance. We maintain a fundamental boundary: publisher transaction data, customer identity details, and API payload information executed through agentic handoffs are strictly isolated. We do not use proprietary third-party transaction data, backend API logs, or enterprise partner telemetry to train our first-party ad models, optimize our own commercial services, or compete directly with partners. In Google Cloud and Vertex AI, as well as in our Search platform interfaces, we adhere to strict enterprise privacy standards, including customer-managed encryption keys, zero-data-retention options for model prompts, and third-party security audits. Furthermore, our system acts as a zero-knowledge broker for sensitive transactions. The AI model facilitates the connection, but the actual data exchange occurs end-to-end between the user and the publisher's secure infrastructure. Our business model succeeds when your platform grows, not by extracting proprietary insights from your operational loop. To ensure complete alignment, would you be interested in reviewing our standard Enterprise Privacy Framework and third-party audit reports to verify how we safeguard partner data isolation?

How do you define "zero-knowledge broker" in the context of your AI model facilitating connections and what specific technologies or protocols are in place to ensure this claim?

In this context, acting as a "zero-knowledge broker" means Google’s orchestration layer routes operational requests and structured payloads without decrypting, storing, or inspecting the underlying payload contents or user PII. To enforce this, we deploy three core technical mechanisms: 1. **Confidential Computing Environments:** Agentic routing and API translation execute within hardware-isolated Trusted Execution Environments (TEEs) using AMD SEV-SNP and Intel TDX. Encryption keys are managed externally via Customer-Managed Encryption Keys (CMEK), ensuring memory contents remain invisible even to root host processes. 2. **Secure Tokenized Handshakes:** Instead of transferring sensitive user credentials or full order payloads through our models, Gemini generates short-lived, single-use OAuth2 tokens and cryptographic hashes. The final transaction payload is passed directly via client-side or point-to-point encrypted API calls between the user's browser and your backend. 3. **Ephemeral Epistemic Processing:** Model context windows operating on transactional data are non-persistent. Once the intent card is rendered and the API token is dispatched, the in-memory context state is instantly purged. No logs, prompts, or payload metadata are retained or routed to training pipelines. These protocols ensure