Google AI Mode Updates – Shopping, Deep Research, Agents, and More
The updates from Google about AI Mode came fast and furious in the last month. Here are the highlights:
- Google AI Pro and AI Ultra subscribers are getting the first look at Gemini 2.5 and Deep Research with Gemini 2.5 in AI Mode. Deep Research with Gemini 2.5 helps “you save hours by issuing hundreds of searches, reasoning across disparate pieces of information and crafting a comprehensive, fully cited report in minutes.”
- Agentic update #1: AI-powered calling to local businesses is available in Search. When a user searches for a local service, they have the option to have Google “call businesses to get pricing and availability information on your behalf — without you needing to pick up the phone.” Just tap the blue “Have AI check pricing” button to set up an agent that will check prices (or availability) on your behalf and text or email you the options it finds.
- “Ask questions about images” is now available on desktop, and the ability to analyze a PDF will be rolled out soon.
- Use Canvas to organize a plan over multiple sessions in a dynamic side panel that updates as you go. You can find Canvas as an experiment in Google Labs to use while in AI Mode.
- Get real-time help with Search Live. Using Google Lens in AI Mode, you can have a conversation with the app about a live video you are shooting. One ROI SEO team member recently used this to troubleshoot a problem with her freezer by having an in-real-time conversation with AI mode about a noise it was making.
- Try on a new outfit without going to the store. Google’s new virtual try-on will let you upload a photo of yourself to show how a new dress, top, or bottom might look on you before you purchase it online.
- Once you’ve found that perfect shirt, in the U.S. you’ll be able to hit “track price,” enter your preferences, and be notified when it fits your budget.
- Bring a vision for your room or a special outfit to life using Google’s Vision Match technology. Just describe what you’re dreaming of, and Google will “generate a range of visual options for your query and use the 50 billion products in the Shopping Graph to show product listings for visual matches we think you’ll like.”
- Agentic update#2: Just last week, AI Mode in English expanded to 180 countries, and is in the process of adding more agentic features, like the ability to reserve a table at a restaurant and find the right event ticket to match your preferences and budget.
If all these new features and experiments are making your head spin, you’re not alone. Google recently explained AI Mode Q4 opportunities to advertisers, including targeting ads in AI Mode based on the context of the conversation, not just the keywords in a user’s query. What does this mean for advertisers that use Performance Max or AI Max for Search?
- Good feed hygiene is key to standing out among the competition.
- Button up your SEO and make your content as helpful as possible. (Ask us about AI SEO/GEO!)
Your SEO and product feed must be top notch to have a good shot at being shown to AI Mode users in Q4.
Bing Says: Follow SEO Guidelines to Optimize for AI Search
We’ve been advising clients to shore up their SEO foundation to rank better in Google’s AI-powered search results since the early appearance of AI Overviews (at the time, they were in beta and called “Search Generative Experiences”). Now Bing’s Search Product Manager, Fabrice Canel, is backing this recommendation. In a recent blog post, Canel and his team shared practical advice for improving content discoverability in AI search – these tips aren’t brand new concepts, they’re normal SEO tactics.
Recommended techniques:
- Update your XML sitemap often.
- Use the Lastmod field in that sitemap (indicates the date a page was last updated).
- Use the IndexNow feature (helps get your content indexed faster).
While these callouts were Bing specific, we continue to emphasize the importance of checking all the technical SEO boxes if you want to appear in AI-powered search results across engines.
Cloudflare blocks Perplexity from crawling websites on account of using stealth, undeclared crawlers; Perplexity denies claim
Cloudflare is alleging that Perplexity is using deceptive, stealth crawlers to circumvent crawl exclusions in sites robots.txt files. After receiving reports that Perplexity was ignoring their crawl parameters from some of their users, Cloudflare investigated by creating net-new websites, then querying Perplexity about the content on those sites, and asking about the permissions of their robots.txt files. Cloudflare concluded that Perplexity was using undeclared user agents and rotating IP addresses to skirt around the robots.txt file. In response, Cloudflare has delisted Perplexity as a verified bot, and blocked the stealth crawling.
An example of how Perplexity allegedly skirted robots.txt rules. Graphic source: Cloudflare.com
In response, Perplexity published a blog post lambasting Cloudflare’s claims and asserting that Cloudflare manufactured this as a publicity stunt. Perplexity says that AI bots behave differently than Googlebot and other traditional search engines, and Cloudflare is failing to recognize the difference between their bot and malicious scrapers.
SEO Takeaway: If you want Perplexity to crawl your site and you use Cloudflare, first confirm that Cloudflare is currently blocking Perplexity. If they are, you can allow Perplexity’s bots in your Cloudflare dashboard.
Will AI-Generated Images Harm Your Website?
In short, Gary Illyes from Google says “no;” however, Search Engine Journal reminds us to make sure the content we generate and post will “resonate with users.” If you were penalized during the March 2024 Core Update due to spammy scaled content abuse, you may be extra cautious with this. But it seems that AI-generated images aren’t the same thing – as long as you use them appropriately without intentionally trying to game the system, of course.
Google Reaffirms Safety Commitments for Kids & Teens
Google is rolling out age assurance – an email that’s sent out when Google tools think a user is likely under 18 – without collecting additional data or sharing user information. The goal is to recognize kids and teens in an effort to better protect them. When identified, these users will receive “age-appropriate” experiences, like reminders to take breaks and go to bed, disabled timelines in Maps, and increased privacy.
Microsoft Copilot Uses Google Business Profiles
Search for the nearest coffee shop, and you’ll find something more interesting (to an SEO) than the new spot for java. Our team has confirmed that Microsoft’s Copilot (formerly Bing Chat) appears to be using Google Maps, reviews, and business profiles to answer local search queries.
Clicking on the map will take you to a Bing maps page, but upon deeper inspection, the map appears to be built with a Google map. The reviews on Copilot and on Google also match, which means that if you want to show up in local search in Copilot, you need to ensure your Google business profile is up to date!
ICYMI: Recent SEO Blog Posts
Why Core Brand Messaging is Your New Superpower in the Age of AI Search
SEO + GEO: Creating a Resilient Strategy in the New World of AI
