Even before Marissa Mayer became the poster-person for one of the worst ways to use AI, I had downloaded her app Dazzle and was planning to use it. There's a strong idea behind it. The execution? Dazzlingly dumb.

I was already on Mayer's list — well, her company's newsletter blast. (I don't know the former Google exec and Yahoo CEO personally.) I was a sporadic user of her Sunshine Contacts app that was a solid stab at organizing address books, even if I didn't stick with it. She dissolved that in September 2025, and Dazzle was born.

The premise: Everyone is constantly taking photos on their phones, and those photos are a goldmine of data relevant to each user, so Dazzle can proactively give you recommendations based on your archive.

That data is also a privacy nightmare that could make Flock cameras seem as tame as baby monitors, but let's put that aside for the moment. I can hear you thinking, "Better you than me!" about downloading this app. You’re right.

A core feature of Dazzle is its “Get It Done” tab where it alerts you daily with recommendations for what it can do based on your latest photos. So, I went to see what it wants me to get done.

First, it showed a photo of groceries delivered in Amazon bags sitting on my counter. It asked, "Would you like to let someone know the groceries and packages are here?"

I'm divorced. The only other person who appears regularly in my photos at home is my middle-school-aged daughter. I responded to the app, "I brought them into my apartment. No other adult lives here. Do you think the packages are sentient beings?" It responded, "No, definitely not." Whew. For what it's worth, my 12-year-old gets very excited about groceries, and the photo was for her sake since she wasn't home when the groceries arrived. I get very excited about groceries too.

The next recommendation stemmed from a screenshot I saved of a pumpkin-carving design. It offered to find a printable stencil of it or to share the design with my daughter. That might be helpful except that by the time it offered this, there were already photos on my phone of the pumpkins my daughter and I each carved. Too little, too late, Dazzle.

Then there were screenshots from a dating app. It said, "Since she is from Chicago and went to University of Chicago, I can help you draft a few personalized opening messages based on her prompts or Chicago roots." First off, unlike Ms. Mayer, I don't need AI for personal interactions. I don’t think it’s attractive to sound like a bot when you’re trying to woo someone. Second, I never save others' profile screenshots, except this profile was clearly fake, and I was using Gemini to help me understand all the fake tells in it so I could properly report it. My screenshots were promptly deleted.

Next photo: me holding fun-size 100 Grand and Butterfinger candy bars, the best one-two punch Halloween has to offer, no debating allowed. Dazzle said, "Would you like me to look up the calories and nutrition facts for these fun-size 100 Grand and Butterfinger candy bars?" My response wasn't subtle: "HELL NO! This is the last thing I want to know while I'm enjoying these!" It keeps doing this, by the way. I made a gorgeous rack of ribs for lunch, and it made the same offer. There is almost a perfect inverse correlation between food that I take pictures of and food that I’d want to know the nutritional info about.

It did have one good idea buried among all the distractions. I got myself a turntable as a birthday present, and it recommended buying a certain 5-1 vinyl record cleaning kit on Amazon, with a link to the purchase. I ordered it right away.

The Ideas tab is the other main feature, full of event ideas, product recommendations, and dining options. The problem here is that the recommendations are mostly all things I either don't want or don't need. It's commercial without being helpful. The one dangerous discovery: The New York Times has a store full of word game merch. Would I look more handsome in a Genius or Queen Bee sweatshirt?

As for where Dazzle's heading, there are two fundamental problems even if the algorithms wind up working better:

1) Privacy. Flock 2.0. Flock is public, this is private. The ick factor is through the roof, more for the worst-case hypothetical scenarios than for how it currently works.

2) This is a feature of other apps. Agentic apps like Instinct already have access to photos. You don't think Muse, built on Facebook and Instagram, won't analyze your photos if you give it permission? What about Google, which has Google Photos plus all the Android devices? Apple has some AI features built into its Photos app and will, of course, roll out more. Dazzle feels like a series of features, not a product.

I'm writing this while watching the third and current season of "Nobody Wants This.” Dazzle should be rebranded with the series name.

Tech-savvy people will use a more sophisticated agentic platform. Most others will shudder at the privacy implications and avoid it. Nobody wants this, and nobody needs this. I may not be climbing telephone poles to sabotage Flock cameras, but I will at least delete my Dazzle account.

— David Berkowitz, Chief Community Officer, Marketecture Media

Who: Retail Marketers, E-commerce Leads, Brand Marketers, Measurement Teams

What: Visitors arriving at retail sites from AI sources converted at a 90% higher rate than other traffic in September and generated 83% more revenue per visit, according to Adobe's AI Traffic Trends Report, reported by eMarketer. AI-referred shoppers also spent 73% more time on site, viewed 25% more pages, and added to cart at a 53% higher rate, while AI traffic to U.S. retail sites jumped 71% year over year. Adobe expects the revenue-per-visit advantage to double during the holiday season. The catch: Attribution is still broken. A PYMNTS study found no surveyed merchant could consistently connect AI-driven visits to purchases, and retailers like Home Depot are restricting agents until they prove they're tied to a real shopper.

Why it matters: The GEO debate just got a P&L: AI referrals are no longer a curiosity; they're the highest-value traffic a retailer gets. But the same data shows the measurement infrastructure hasn't caught up. Brands are being told to invest in visibility they can't yet attribute, which is exactly the kind of gap that breeds both opportunity and snake oil.

(eMarketer)

Who: CPG Marketers, Brand Marketers, E-commerce Leads, Agency Strategists

What: 80% of U.S. consumers already use AI, and nearly half of those use it to shop. But 59% of CPG companies report no meaningful agentic AI capabilities, according to BCG research published Oct. 30. The readiness gap carries real money: One global CPG assessment found 20% to 30% of online revenue could be at risk as discovery shifts to AI search, while companies piloting GEO are seeing 1% to 5% incremental revenue and those testing owned agents and ecosystem integration are driving 3% to 10% growth, with first-party data from owned agents lifting digital media ROI by up to 50%. BCG expects AI-driven traffic to overtake organic search in the U.S. as early as 2028.

Why it matters: The shoppers are already there, the brands mostly aren't. The 1%-5% GEO revenue figure is one of the first credible attempts to size what AI visibility is actually worth. Expect it to be quoted in every agency pitch by January.

(BCG)

Who: Agency Executives, SEO/GEO Leads, Brand Marketers, MarTech Buyers

What: With a glut of answer-engine-optimization platforms pitching brands, agencies are building evaluation frameworks rather than committing to one tool, executives from VML and Omnicom told Marketing Brew. VML's chief discoverability officer Heather Physioc prioritizes platforms that measure visibility across multiple LLMs and track trends over time while warning that none of today's proprietary "AI brand scores" is a North Star metric. Omnicom's head of AI platforms, Dan Toplitt, scrutinizes data collection methodology, like whether a platform uses APIs or headless browsers, and refuses to make strategy changes off three-day data refreshes. Both expect standardized AEO measurement to eventually emerge.

Why it matters: This offers a practical buyer's guide from the people actually spending client money on these tools, and a warning shot at vendor marketing. The methodology question (API versus headless browser) is the detail that separates serious evaluation from dashboard shopping, and "don't make knee-jerk decisions off daily refreshes" deserves to be pinned above every CMO's desk.

(Marketing Brew)

Liking AI Brief? The Refresh is a weekly recap of all the most important advertising news, written by AdTechGod.

Thanks for reading this week’s AI BRIEF by Marketecture. 

Marketecture Media is committed to bringing you the latest news, opinions, and community updates. 

If you are interested in advertising on this newsletter, our podcasts, The Advertising Forum, or in our Slack community, you can contact us here.