Milena Traikovich stands at the forefront of the rapidly evolving intersection between artificial intelligence and consumer behavior. As a seasoned expert in demand generation and digital ecosystems, she specializes in transforming complex technological shifts into actionable strategies for businesses aiming to capture high-quality leads. With the recent unveiling of Google’s latest AI-driven search and commerce tools, Milena provides a crucial bridge between advanced technical protocols and the real-world experiences of both merchants and shoppers.
This conversation explores the transition to Gemini 3.5 Flash, the emergence of autonomous agents that can make phone calls and book services, and the logistical hurdles of the Universal Commerce Protocol. We also delve into the security measures behind the Agent Payments Protocol and the massive infrastructure required to watermark over 100 billion pieces of digital media to ensure transparency.
With Gemini 3.5 Flash now serving as the default for AI-driven searches, how does the ability to input entire files or browser tabs change the way users interact with information? Could you explain the technical shifts required to maintain conversational context while query volumes continue to scale rapidly?
The shift to a multi-modal input system, where users can simply drop in a video, a massive data file, or even an active Chrome tab, represents a move away from “keyword hunting” toward true “contextual understanding.” It changes the user behavior from asking a question to providing a problem set for the AI to solve. Technically, managing this requires an incredible infrastructure—AI Mode queries have already more than doubled every quarter since launch, and we are now seeing over one billion monthly users. To keep up, the system uses a redesigned search box that expands dynamically, allowing the AI to maintain a continuous thread of context as a user moves from an initial query into a deeper AI Mode conversation. This ensures that when you ask a follow-up question, the AI still “remembers” the data from the 50-page PDF or the specific video timestamp you provided earlier.
Specialized agents are now monitoring real-time data for high-stakes categories like finance and home services. How should small businesses in sectors like pet care or repair prepare for an AI calling them on a customer’s behalf, and what steps ensure these bookings remain accurate and reliable?
Small business owners need to brace for a summer where their phones might start ringing with an AI on the other end of the line. Google is expanding these agentic booking features to local services like pet care and home repair specifically to bridge the gap between a user’s specific requirements and a merchant’s availability. For a business, preparation means ensuring their online presence, pricing, and scheduling data are structured and accessible, as these agents scan real-time data to find the best match. The reliability comes from the fact that the Search results return specific pricing and direct links to complete bookings, meaning the AI isn’t just “guessing”—it’s acting on verified availability parameters. It’s an emotional shift for a business owner to realize their first interaction with a new client might be fully automated, but it’s a massive opportunity for lead generation.
The Universal Commerce Protocol allows for a single shopping experience across disparate platforms like video and email. What are the logistical challenges in synchronizing inventory and price history from multiple merchants, and how does the system effectively flag compatibility issues between products from different retailers?
The primary logistical hurdle is creating a “common language” for commerce that can track a product’s journey from a YouTube ad to a Gmail checkout. By utilizing Gemini models to monitor the Universal Cart, the system can actually provide a history of price drops and stock alerts across 200 countries and territories. What’s truly impressive is the logic required to flag compatibility; for example, if you are building a computer and add a motherboard from one merchant and a CPU from another, the system cross-references technical specs to warn you of a mismatch. It aggregates your loyalty information and payment benefits from Google Wallet in real-time, ensuring that even if you are buying from three different brands, the checkout feels like a single, cohesive event. This level of synchronization requires a massive backend that keeps the brand as the merchant of record while Google handles the heavy lifting of the data reconciliation.
Establishing digital mandates and spending limits is central to the new Agent Payments Protocol. How do these verifiable links between the user and the payment processor prevent unauthorized spending, and what specific workflows will be visible to users as these automated tools begin handling purchases?
The Agent Payments Protocol, or AP2, is built to solve the “trust gap” that occurs when you give an AI the keys to your wallet. It works by creating a digital mandate—essentially a recorded set of instructions that defines exactly what the agent is allowed to do. Users will see a workflow where they pre-define the brand, the specific product, and a strict spending limit before the agent is even authorized to initiate a transaction. This creates a verifiable link between the user, the merchant, and the payment processor, providing a transparent audit trail for both purchases and returns. When these tools roll out, starting with Gemini Spark, the user remains the ultimate gatekeeper, seeing a shared record of every action the agent takes to ensure no “hallucination” leads to an accidental or unauthorized charge.
Scaling watermarking technology to over 100 billion pieces of media suggests a massive infrastructure for digital transparency. How will the integration of content credentials into smartphone cameras and social feeds impact consumer trust, and what are the primary hurdles in getting other major global platforms to adopt these standards?
We are entering an era where seeing is no longer believing, which is why scaling SynthID to over 100 billion images and 60,000 years of audio is so vital. By embedding signals directly into AI-generated content, and integrating C2PA Content Credentials into hardware like the Pixel 10 and social feeds like Instagram, we are giving consumers a “nutrition label” for digital media. The impact on trust is profound—users will be able to use tools like Google Lens or Circle to Search to instantly verify if an image is an unaltered original or a synthetic creation. The main hurdle for global adoption is the “walled garden” problem; however, seeing major players like Meta, OpenAI, and NVIDIA commit to using these watermarking standards suggests we are reaching a tipping point. It requires a unified front where every platform, from a niche audio generator to a global social network, agrees on the same digital fingerprinting language.
What is your forecast for the future of AI-integrated search and commerce?
My forecast is that we are moving toward a “zero-friction” economy where the distance between a desire and a completed transaction is almost entirely collapsed by personal intelligence. We will soon see a world where your AI doesn’t just find you a product, but proactively manages your life—coordinating your Google Calendar with a local delivery service and using your Agent Payments Protocol to buy a gift before you even realize you’ve forgotten an anniversary. Within the next two years, I expect generative user interfaces to become the norm, meaning your search results won’t be a list of links, but a custom-built dashboard with tables, trackers, and simulations tailored specifically to your current intent. Commerce will no longer be a destination you visit, but a persistent, helpful layer that exists across every digital surface you touch.
