The Deepdive
Join Allen and Ida as they dive deep into the world of tech, unpacking the latest trends, innovations, and disruptions in an engaging, thought-provoking conversation. Whether you’re a tech enthusiast or just curious about how technology shapes our world, The Deepdive is your go-to podcast for insightful analysis and passionate discussion.
Tune in for fresh perspectives, dynamic debates, and the tech talk you didn’t know you needed!
The Deepdive
Ruthless Shopping - How AI Superagents Rewrite E-commerce
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
A quiet revolution is changing how we buy everyday things: agents that shop for us. We explore the rise of agentic commerce, where AI systems parse structured data across retailers, optimize for your parameters, and complete purchases with near-zero friction. From a late-night fever fix to a one-text taco dinner, we show how the cognitive burden of shopping shifts from your brain to software—and why that shift could be the biggest retail disruption since e-commerce.
We break down what makes cross-platform superagents different from walled-garden tools, spotlight Perplexity Shopping’s intent-aware recommendations, and examine instant buy flows that compress checkout into chat. Then we dig into memory and personalization: how assistants learn your dietary needs, brand preferences, fitness goals, and pain points to deliver precise answers in seconds. The payoff is speed and savings; the trade-off is data. We map the split between price-sensitive pragmatists who rush toward automation and privacy skeptics who demand control, encryption, and easy toggles to disable memory and tracking.
It’s not all smooth. Early agents stumble on stale inventory and brittle checkout funnels, sometimes requiring human oversight. Still, the direction is clear—and retailers feel it. We explain how bot logic drives price compression, erodes margins, and commoditizes brands that can’t prove value in structured data. The path forward is becoming agent-preferred: clean, real-time feeds, dynamic pricing, reliable availability, transparent fees, and fast, predictable fulfillment. We close with a provocative question about values and optimization: if your agent finds the best price from a brand you reject on principle, how much are you willing to pay for ethical alignment?
If this conversation sparked ideas or concerns, follow the show, share it with a friend, and leave a quick review telling us where you draw the line with AI shopping.
Leave your thoughts in the comments and subscribe for more tech updates and reviews.
Agentic commerce is here. And uh, according to the sources we've got in front of us, it's being called the single most disruptive shift in retail since since the original rise of e-commerce itself. I mean, that is not a small claim. So our mission today, I think, is to figure out why. What happens when that fundamental relationship between you and a retailer just breaks? What happens when you realize you don't have to spend your own mental energy scrolling for necessities anymore? What happens when your agent shops and you well, you just approve the transaction?
Allan:Yeah. And that's the core definition, right?
Ida:Trevor Burrus, Jr. That's it.
Allan:These AI systems act on your behalf. They make purchases either autonomously or you know, semi-autonomously.
Ida:Trevor Burrus, Jr.: Trevor Burrus, Jr.: And then they just execute the transaction.
Allan:All with uh very limited human oversight.
Ida:Aaron Powell And this isn't some far-off theoretical thing in a white paper. Trevor Burrus, Jr.
Allan:Not at all. The consumer expectation data is it's immediate. We're seeing numbers that show 60% of consumers anticipate using AI agents for shopping within the next 12 months.
Ida:60 percent. Within a year.
Allan:It's happening fast. And that enthusiasm, it's being driven by something really specific.
Ida:Aaron Powell It's not just about speed, is it? It's not just saving 10 minutes on your weekly grocery run.
Allan:No, it's the belief that the uh agent will genuinely be better at the job than you are. Look at the data on pricing.
Ida:Okay.
Allan:74% of consumers believe these agents will be effective at finding the absolute right price. That's the real motivator, the savings combined with the convenience.
Ida:Aaron Powell And this is where the disruption gets really fascinating for the retailers themselves. We aren't just talking about, say, Amazon embedding a slightly smarter Alexa into its own ecosystem.
Allan:Trevor Burrus Right. To get you to reorder paper towel.
Ida:Exactly. That's just vertical integration.
Allan:Aaron Ross Powell What we're talking about is the rise of these cross-platform superagents.
Ida:Aaron Powell Okay, so what does that mean?
Allan:Aaron Powell These are sophisticated tools that sit squarely between the consumer and well, all retailers. Yeah. They don't have any loyalty to Amazon or Target or your favorite little brand.
Ida:Trevor Burrus They just shop.
Allan:They choose products based purely on bat logic, on structured data. They prioritize your parameters above everything else. It's like having a fully informed, totally ruthless personal shopper who works for you 24-7.
Ida:Aaron Powell I think we need to spend a minute just appreciating the convenience here because it's not just about efficiency, it's about outsourcing the tedious, repetitive mental load of modern life. Our sources give us two great, really relatable scenarios. The first one is the surprise fever story.
Allan:Oh, yeah. This one's great.
Ida:It's 10 at night, your kid spikes a fever, and your medicine cabinet is, of course, empty. Instead of scrambling, you just whisper to your agent, I need kids to see the minophil right now.
Allan:And that one single command just replaced what, maybe 30 mental steps for a person? Oh, easily. You're bypassing, comparing brands, checking local inventory, deciding if you should braze the pharmacy, or pay for delivery. All of that. The agent just instantly cross-references availability and cost. And in that scenario, you just get a notification available at CVS for curbside pickup in 12 minutes, or delivered in 40 minutes for $4 more. You just tap deliver.
Ida:And you've completed a crisis purchase while staying focused on your kid. You didn't worry about the store, the brand, the dosage. The problem was just solved.
Allan:It's cognitive relief.
Ida:Totally. Or take the taco night scenario. You just text your agent, add ingredients for chicken tacos for four people tonight.
Allan:The agent pulls a recipe you like, it compares the price of chicken, cheese, tortillas across, let's say, three local grocers.
Ida:Whole Foods, Kroger.
Allan:Instacart's Marketplace, yeah. And it schedules the delivery for 5 p.m. All of that in maybe 20 seconds.
Ida:You didn't open a single app, you didn't check three different digital flyers, you didn't type in a credit card number.
Allan:Nothing.
Ida:So what's actually underpinning this? I mean, this level of ruthless efficiency, it's important to get that agents don't shop like we do. We browse, we look at glossy pictures.
Allan:We read customer reviews.
Ida:Yeah, how do they actually decide?
Allan:They ignore all of that. They ignore the glossy website, they bypass all the content. They are only interested in one thing extracting highly structured data.
Ida:Structured data.
Allan:To make parameter-based decisions. So when you say chicken, they are not looking at the farm to table story on the website. No. They're looking at price per pound, inventory level, and the fulfillment window. That's it.
Ida:So structured data means the agent isn't reading the marketing copy. It's reading what? Like a spreadsheet version of the product.
Allan:Aaron Powell That is a perfect way to put it. Yes. They're reading APIs, inventory feeds, pricing updates. And this is why consumers have such high expectations. They want the agent to excel at finding the right product. That's a 68% expectation. The right price, 74%, and the right time, 72%. U teen tasks just vanish. Imagine your agent connected to a smart laundry dispenser.
Ida:Right. It pings you. You're down to two loads left. It finds the lowest price for your preferred detergent.
Allan:And it delivers it with a one-tap approval from me.
Ida:And just like that, shopping becomes this quiet background process you don't even think about. All right, let's get into the specific disruptive players, these super agents. Because while platforms like, say, Instacart's cart builder are smart, they're still basically inside a specific company's walls.
Allan:They're embedded, yeah.
Ida:Yeah.
Allan:The big challenge comes from companies that are truly cross-retailer.
Ida:And the best case study for this right now is perplexity shopping.
Allan:They represent a fundamental shift in how a transaction even begins. They're not just a search engine with a shopping tab. They use natural language processing to understand the uh the intent behind complicated human questions.
Ida:So instead of me typing wireless headphones black, I'm asking something more human, like, what are the best wireless headphones for travel that have noise cancellation but won't fall out when I jog through the airport?
Allan:Exactly. That's a messy human question.
Ida:It is.
Allan:And Perplexity's Genius is delivering these curated, synthesized product recommendations based on that nuanced intent. It pulls data from reviews, product listings, expert summaries, and it gives you an answer that looks more like a research report than a product page.
Ida:And this is the critical difference, right? From traditional platforms that are just dominated by SEO and paid placements. Perplexity claims to prioritize relevance over promotion.
Allan:That's the claim. Yeah. They aim to surface the highest quality information for you, the shopper, not just the item that paid the most to be seen first.
Ida:This is where we need to kind of pause and think critically, I feel, because on the surface, this builds stronger consumer trust.
Allan:Sure. It feels transparent. They cite their sources, they summarize credible information, they explain why a product is recommended.
Ida:Are we just trading one black box, you know, the opaque world of uh SEO algorithms for a new black box? Which is the opaque world of bot logic.
Allan:That's the question. We are putting a huge amount of trust in the agent's definition of what best even means.
Ida:That's a great point. But for the average person, the convenience is so immediate. Perplexity has partnered with PayPal for an instant buy feature.
Allan:And that's all about reducing friction.
Ida:Totally. We all hate those frustrating multi-step checkouts. Here you complete the whole transaction right inside the chat window.
Allan:It streamlines that path to purchase immediately. Now, right now, that feature is limited to a small list of partners, I think five brands.
Ida:Yeah, like Abercrombie and Fitch, Ashley Furniture, a few others.
Allan:But it shows the direction of travel. Minimize the checkout friction wherever you possibly can.
Ida:And crucially, the agent is just the matchmaker in all this. The retailer is still the official seller.
Allan:Right. So if you buy a sofa through the perplexity chat, the retailer is still on the hook for shipping, returns, customer support.
Ida:Aaron Powell The agent just handled the research and the seamless transaction part.
Allan:Exactly.
Ida:Okay, now we get to the feature that makes this revolutionary, but also maybe slightly unnerving. And that's the agent's memory. An agent that knows you better than you know yourself.
Allan:Aaron Powell Yeah. And the development of features like Perplexity's Comet Assistant, this is what's driving that extreme personalization. The system automatically synthesizes key details and uh structured preferences over time.
Ida:It's storing my favorite brands, my dietary needs. Trevor Burrus, Jr.
Allan:Whether you prefer organic, your health and fitness goals, all based on your past searches.
Ida:Aaron Powell So when I ask for a recommendation, the agent isn't starting from scratch every time. It's building this cumulative profile of me.
Allan:Precisely. The classic example is running shoes. If you ask, recommend new running shoes for me, the assistant might already know that you're training for a marathon.
Ida:Or that I recently searched for info on a nagging foot injury.
Allan:Exactly. It understands the underlying context of your life. And so it provides a much more precise answer than any human assistant could in 20 seconds.
Ida:This level of personalization is, I mean, it's incredible, but it naturally leads to that tension we touched on before: convenience versus data control. And the market is not a monolith on this.
Allan:Not at all. The sources break the consumers down into really clear segments. You have the price-sensitive pragmatists, they're about 35% of the market.
Ida:And they are rushing toward this technology.
Allan:Oh, yeah. They only care about the savings and the efficiency. They are more than happy to trade data for a lower price.
Ida:But then you have the privacy conscious skeptics. That's about 30% of the market.
Allan:And these are the ones asking the fundamental questions. Wait a minute. You're collecting and tracking my consumption habits, my location, my medical history in this memory feature.
Ida:They place a very high value on data security.
Allan:And on retaining control over the final decision. And that control is key to winning them over. These systems are designed so you, the user, remain in the driver's seat. So you can just turn it off? You can easily turn the memory feature off entirely. And if you browse privately, like in an incognito mode, the memory and search history are automatically disabled.
Ida:And the data is encrypted.
Allan:It has to be. System needs to respect that segmentation, or the skeptics just won't adopt it. Period.
Ida:So we've painted this picture of seamless, effortless purchasing. But before we all, you know, hand over our financial lives to a machine, we need a reality check. This technology is still evolving, and the early days are always messy.
Allan:Oh, they are definitely messy. The early tests of Perplexity Shopping Agent really highlight the challenges of true autonomy. We are still very much in the beta phase.
Ida:So give us an example of a bot going bad.
Allan:Well, the sources cite this specific difficulty with a simple routine task, just buying a tube of crest toothpaste.
Ida:Okay, what happened?
Allan:The process took hours, and sometimes the agent just failed outright.
Ida:Why?
Allan:Because the data it was scraping from the retailer's website was stale. The system thought the product was in stock, but when it actually tried to execute the purchase, it had been sold out at Walmart for hours.
Ida:So the agents are still making these basic, frustrating, almost human-like mistakes because they're relying on static information instead of real-time inventory.
Allan:Precisely. You are paying the agent, and the agent then attempts to buy it on your behalf, but that task might not be instantaneous. A perplexity spokesperson even acknowledged that they currently require human oversight, providing occasional support.
Ida:Ah. So there's a person behind the curtain sometimes?
Allan:To make sure the transactions are timely and correct. It's not fully autonomous yet.
Ida:But despite these technical hiccups, the entire retail industry's Wall, they're sweating. Because they know where this bot logic is ultimately headed.
Allan:They do.
Ida:And for retailers who are unprepared for this shift, the financial consequences are severe. We're talking about potential EBIT erosion of up to 500 basis points.
Allan:500 basis points. That is a devastating loss of operating profit margins.
Ida:It's huge.
Allan:And that collapse is due to two related things. First, the agent's logic. It just optimizes relentlessly for the lowest price, which forces price compression across the entire market.
Ida:And second, brand commoditization. My fancy logo and my expensive marketing campaign mean absolutely nothing to an agent that only reads price, ingredient list, and delivery window.
Allan:Exactly. Brand loyalty, brand perception. It matters so much less than the structured data points. So now retailers have to become agent preferred.
Ida:What does that even mean?
Allan:It means overhauling their entire digital infrastructure. They have to make sure the pricing is dynamic, their inventory feeds are clean and readable by algorithms, and their fulfillment speed and reliability are just superior. They have to win the battle of the bots to survive.
Ida:So let's bring this all back home. Agentic Commerce is fundamentally transforming how we get the things we need. It's prioritizing maximum efficiency and price transparency above all else. The convenience getting chicken tacos delivered without opening a single app is enormous, but that convenience is directly proportional to how much autonomy and how much data we hand over to our AI assistant.
Allan:Yeah, and we've looked at the consumers driving this. You've got the pre-sensitive pragmatists rushing toward savings, and you've got the privacy conscious skeptics who are demanding oversight. The core tension we all face is balancing that absolute relief of outsourcing mental labor.
Ida:With maintaining critical data control.
Allan:Exactly. And this transition into a world governed by bot logic leads us to our final provocative thought. It's something for you to mull over as your new agent starts making choices on your behalf.
Ida:If your AI agent is designed to find the absolute best price and the most efficient fulfillment, and the retailer that consistently offers that optimal service is a company, a brand that you actively dislike or distrust for reasons outside of product quality.
Allan:Ethical reasons, personal reasons.
Ida:So you program your agent to override its own core logical optimization just for the sake of your feelings. In this new economy, mediated by ruthless efficiency, what is the exact economic value of brand spite?
