Automated Suggestions in Shopping Apps

One of the biggest challenges when building real-time recommenders? Most people think it’s just about making accurate predictions... But the real challenge is narrowing down from millions of potential item candidates to just a few personalized recommendations. And it must happen in less than a second. This is where the 4-stage recommender architecture comes into play... It's a scalable framework used by companies like NVIDIA and YouTube to personalize recommendations in real-time. I want to walk you through how we can apply this architecture to a real-world use case: H&M’s fashion recommendation engine. The problem: At H&M, the goal is to recommend fashion items to millions of customers based on their browsing and shopping history. For example, if a customer searches for T-shirts, the recommender should automatically prioritize personalized T-shirt suggestions. But how can this be done in real-time? 4-Stage Recommender Architecture Here’s how H&M uses the 4-stage architecture to make this happen: Stage 1: Candidate Generation When customers surf the H&M app, their ID and date are sent to the recommender system. The Customer Query Model computes a customer embedding based on these inputs. This embedding is then compared to a vector index of all H&M’s fashion items, which helps narrow millions of items to a coarse list of hundreds of relevant articles. Stage 2: Filtering Next, using a Bloom filter, we filter out items the customer has already seen or purchased. This step reduces the list of candidates to a more focused set, eliminating unnecessary redundancies. Stage 3: Ranking The remaining items are ranked based on their relevance to the customer. At this stage, the Hopsworks feature store provides features in real-time from its online store describing the item and the customer relationship. This enables a CatBoost model to score the list of hundreds of items more accurately in real time. Stage 4: Ordering & Business Logic Finally, the items are ordered based on their relevance scores and any additional business logic (e.g., promotional items or new collections). We reduce the final list to a few dozen highly personalized recommendations. The customer now sees fashion items they are most likely to click on and buy. The entire architecture is powered by Hopsworks, an AI Lakehouse that provides: - A Feature Store that stores the features in an online store accessible for real-time inference. - A Model Registry to manage the query, ranking, and candidate encoder models. - A Serving Layer to deploy the recommender system in production. 🔗 Curious to dive deeper? Check out how we built this step-by-step: → https://lnkd.in/dmvpHPyT

🚨Breaking🚨 Amazon just turned Rufus into an agent. Every AI shopping assistant waits for you to ask. Amazon just stopped waiting. Amazon rolled out 'Scheduled Actions' in Rufus last week. It offers a handful of suggested use cases to get shoppers started, and the one I clicked on for the video below was "Set a birthday reminder with gift ideas." Sounds boring, but it's actually quite a bit shift in capability. Gift discovery is one of the biggest use cases driving people to ChatGPT and other general-purpose LLMs for shopping help. "What should I get my brother-in-law who likes cycling and hates gadgets" is exactly the kind of query Amazon can't afford to lose to OpenAI. Scheduled Actions does two jobs at once: it pushes Amazon further into agentic commerce, and it builds a moat around many of retail's most lucrative, most AI-native categories. Scheduled Actions lets Rufus act on a schedule instead of only responding to live queries. Amazon's other starter suggestions include monthly coffee recommendations, new book alerts, and automated cleaning supply restocks. But the suggested prompts aren't the story. They're training wheels. The real shift is that shoppers can define their own. 🗣️ Remind me to buy running shoes every 18 months. 🗣️ Alert me when the price on [specific product] drops below $X. 🗣️ Send me a new hot sauce every fortnight. 🗣️ Remind me about pet food before it runs out, based on the pack size. 🗣️ Suggest new swimwear before the kids outgrow them. Amazon just handed shoppers a programmable layer over their own commerce life. Why the birthday example is an intersting one: 🛒 Gift-giving is one of the highest-intent, most brand-agnostic queries in commerce. "Gift ideas for mum" doesn't care whose brand ranks, it cares about fit. 🛒 Birthdays are recurring and predictable. Amazon now has a calendar of future purchase intent, not a guess. 🛒 It captures relationship data Amazon didn't own before. Who matters to you, their age, what you bought last year, what they actually liked. 🛒 It moves discovery off Google and ChatGPT and into Amazon's own interface, where the shelf is stocked and checkout is one tap. For brands, this rewrites the brief. If your product isn't in Amazon's logic for specific occasions, usage patterns, replenishment cycles, and relationships, you don't get scheduled. Structured attributes stop being a hygiene factor and become the entire shelf. For retailers, the warning is sharper. Every scheduled action Amazon captures is a future purchase locked in before anyone else gets to pitch. No ad auction. No comparison tab. No competing retailer even in the room - it's definitely time to begin exploring your own conversational assistant. The interesting question isn't which use cases Amazon suggests💡 It's what shoppers will invent once they realise Rufus can just do it for them 🤔

Google just announced a suite of AI-driven shopping tools designed to transform the online retail experience 💥 Here’s what’s coming: 🧠 AI Shopping Mode: A conversational search tool that understands natural language and context. Think: “best dress for a trip to Cannes in May” – and it will deliver results tailored to weather, location, and purpose. 👗 Virtual Try-On 2.0: Shoppers can now upload full-body photos to see how items actually look on them — including drape, fit, and texture across different body types. A game-changer for apparel conversion rates. 💸 Agentic Checkout + Price Alerts: Users can set price targets, and when an item hits the right price, Google will notify them — or even auto-checkout via Google Pay. What this means for eComm marketers: ▪️ SEO and product data quality will become even more critical ▪️ Conversion will increasingly depend on visual assets (UGC, try-ons, real models) ▪️ Brands need to prep for a world where shopping journeys are conversation-first Currently, these features are being rolled out in the U.S. via Google Search Labs, with plans for broader availability in the future.

As an analyst, I was intrigued to read an article about Instacart's innovative "Ask Instacart" feature integrating chatbots and chatgpt, allowing customers to create and refine shopping lists by asking questions like, 'What is a healthy lunch option for my kids?' Ask Instacart then provides potential options based on user's past buying habits and provides recipes and a shopping list once users have selected the option they want to try! This tool not only provides a personalized shopping experience but also offers a gold mine of customer insights that can inform various aspects of a business strategy. Here's what I inferred as an analyst : 1️⃣ Customer Preferences Uncovered: By analyzing the questions and options selected, we can understand what products, recipes, and meal ideas resonate with different customer segments, enabling better product assortment and personalized marketing. 2️⃣ Personalization Opportunities: The tool leverages past buying habits to make recommendations, presenting opportunities to tailor the shopping experience based on individual preferences. 3️⃣ Trend Identification: Tracking the types of questions and preferences expressed through the tool can help identify emerging trends in areas like healthy eating, dietary restrictions, or cuisine preferences, allowing businesses to stay ahead of the curve. 4️⃣ Shopping List Insights: Analyzing the generated shopping lists can reveal common item combinations, complementary products, and opportunities for bundle deals or cross-selling recommendations. 5️⃣ Recipe and Meal Planning: The tool's integration with recipes and meal planning provides valuable insights into customers' cooking habits, preferred ingredients, and meal types, informing content creation and potential partnerships. The "Ask Instacart" tool is a prime example of how innovative technologies can not only enhance the customer experience but also generate valuable data-driven insights that can drive strategic business decisions. A great way to extract meaningful insights from such data sources and translate them into actionable strategies that create value for customers and businesses alike. Article to refer : https://lnkd.in/gAW4A2db #DataAnalytics #CustomerInsights #Innovation #ECommerce #GroceryRetail

I went viral with a playbook for AI-native marketing… and I didn’t even get mention Instant Checkout. So here it is. 📣 OpenAI just rolled out Instant Checkout inside ChatGPT. You can now buy from Etsy (and soon Shopify) without ever leaving the chat. Zero-click commerce is here! If you’re selling online and especially on Etsy or Shopify, here are 10 things you should be doing right now to make sure the AI suggests you and not your competitor: 1/ Nail your product data Titles, descriptions, tags, attributes, all written in natural language that mirrors how people ask questions, not just how search engines crawl. 2/ Build semantic signals Think less about keywords, more about clusters of meaning. Your product copy should cover synonyms, use-cases, problem statements - the way a person might phrase the query. 3/ Strengthen your reviews AI is not going to suggest a store with a shaky reputation. Make reviews part of your growth loop: follow-up emails, UGC incentives, loyalty perks. 4/ Audit your fulfillment Speed, reliability, and refund rates will become inputs to AI trust. If you’re slow to ship or messy on returns, expect to be deprioritized. 5/ Align content with product Your blog, socials, even your “About” page should echo the same phrasing your customers use. The more consistent the signals, the easier it is for AI to “understand” your relevance. 6/ Strengthen domain + brand trust LLMs scrape broadly. Mentions in earned media, creator shout-outs, affiliate placements — these are credibility signals that feed discoverability. 7/ Test conversational queries Literally ask ChatGPT: “What’s the best X on Etsy?” “Where should I buy Y on Shopify?” See what it says. If you’re not there, work backward on why. 8/ Clean up your backend Inventory, pricing, shipping APIs — these need to be rock solid. If the model detects friction, it won’t risk surfacing you in one-click purchase flows. 9/ Lean into affiliates and partnerships They’re still foundational. But now, it’s not just about clicks — it’s about putting your brand into the semantic web the AI references when it makes a recommendation. 10/ Build brand memorability AI doesn’t only pull structured data. It picks up on what humans repeat. Memorable phrasing, distinctive positioning, unique angles — these increase the odds that your brand surfaces in the model’s “memory.” This is the new funnel: not awareness → click → checkout. It’s awareness → AI suggestion → purchase.

Amazon's AI just became your biggest competitor …and your most important customer. Amazon rolled out a feature last week that should have every seller and advertiser paying attention. "Help Me Decide" uses AI to analyze a shopper's recently viewed items and hand-picks 3 products for them: a "best pick," a "budget option," and an "upgrade pick." It provides AI-generated reasons why each product was selected, pulling from reviews, features, and the customer's shopping history. As an Amazon PPC agency working with sellers across many categories, we believe this feature introduces both opportunity and a new priority. ✅ Opportunity: Shoppers are being guided faster to a decision. If your listing is chosen by the AI, that could mean fewer distractions, fewer competing clicks, and essentially a more direct path to purchase. 📌 Priority: The AI picks products based on browsing behavior + listing signals (features, reviews, alignment with shopper’s prior purchases). That means you need to ensure your listing and your PPC strategy tick all the boxes: compelling unique selling proposition, strong review profile, clear feature/benefit language, and pricing that works in the context of “budget vs upgrade”. From a PPC lens, here are three immediate action steps we recommend: ➣ Review your listings and ensure the key differentiators are front-and-centre (because the AI will use them). ➣ Adjust your ad creative and targeting to capture interest before the indecision threshold (i.e., fewer “I’m just browsing” shoppers). ➣ Monitor your metrics and attribution: as this feature rolls out, watch for shifts in how often your ads lead to a listing view vs. how often a listing gets surfaced by AI-driven recommendation. In short: if Amazon is automating more of the decision process, the quality of your listing and how well your PPC strategy supports that listing become more critical than ever. If you’re a seller wondering how to make sure you’re not passed over by “Help Me Decide”, send me a DM.

AI Is Quietly Deciding Which Products Win https://lnkd.in/gtQD2J7Z Not by brand size. Not by ad spend. By data clarity. When shoppers ask AI what to buy, only products with clean, structured information get recommended. At Preo Communications, we’ve watched smaller brands outsell bigger competitors simply because AI could understand their products better. What gives products an edge in AI shopping: Clear titles that state who it’s for and why it matters Complete attributes (size, color, material, use case) Accurate pricing and stock High-quality images AI can interpret AI doesn’t reward effort. It rewards clarity. In AI-driven commerce, the clearest products win the sale.

Amazon just changed the game for voice-activated commerce. When Alexa launched in 2014, I was on Procter & Gamble's Amazon team and we believed voice-activated shopping was the future. But, it never took off. Why? Because the consumer experience was awful. But now? That’s changing. Amazon just announced a major AI upgrade for Alexa. Alexa+ will be powered by generative AI. This will make voice assisted shopping more intuitive, more personal, and more proactive. - SHOPPING WILL BE SHAPED BY DAILY CONVERSATIONS, NOT JUST SEARCHES: If you ask Alexa for dinner ideas, she’ll start to learn your preferences. So later, when you say, “Alexa, add milk to my cart,” she’ll know you prefer oat milk. - RECOMMENDATIONS WILL BE SMARTER AND CONTEXT-AWARE: If you ask Alexa to set a reminder for your kid’s soccer game every Saturday, she’ll remember. So, weeks later, when you say, “Alexa, I need snacks for the team,” she’ll suggest bulk options that fit your preferences. - PERSONALIZATION WILL MOVE BEYOND PAST PURCHASES: Alexa will pick up on lifestyle details like who’s in your household, what activities you do, and even seasonal needs so she can recommend the right products at the right time. - MULTI-STEP PURCHASES WILL FEEL LIKE HAVING A PERSONAL ASSISTANT: Say, “Alexa, we’re going to the beach this weekend. What should I bring?” She’ll ask follow-ups, suggest essentials, and build your cart - handling the details so you don’t have to. - BUYING WILL BE PREDICTIVE, NOT REACTIVE: Instead of waiting for you to run out of coffee, Alexa will anticipate when you’ll need more based on past orders, how often your coffee maker runs, if you’ve got guests coming, and other conversations you’ve had. She’ll give you a heads-up before you even think about reordering. How to Take Action TODAY ✅ For Brands - Test Alexa’s Understanding of Your Brand Ask Alexa about your brand, category, and competitors. What is she recommending? If your products aren’t surfacing correctly, refine your product data and positioning to align with AI-driven discovery. ✅ For Brands – Optimize for Conversational AI Alexa+ doesn’t just pull from listings—it answers real-world questions. Ensure your product content clearly communicates features, benefits, and differentiators in a way that AI can understand and recommend. ✅ For Retailers – If Alexa+ is making shopping seamless and predictive within the Amazon ecosystem, retailers outside of Amazon need to rethink how they compete. The biggest risk is losing shopper mindshare if consumers get used to “letting Alexa handle it.” What do you see as the biggest implications of Alexa having generative AI embedded? How do you think it will this shift impact shopping, marketing, and the way consumers interact with brands? See comments for link to Amazon Alexa+ news article. Jessy Stamates Kiri Masters Retail AI Council Sanjay Parihar Marie Matacia Ashley Rozier Charlie Chappell Ilie Ban Phillip Jackson Paul Acerbi Kristy Click Morgan McAlenney

ChatGPT Launched Shopping Research Today, OpenAI launched Shopping Research within ChatGPT. Shopping Research uses a dedicated version of ChatGPT version of GPT‑5 mini trained specifically for shopping tasks. It uses reinforcement learning to read trusted retail websites, cite reliable sources, and deliver high-quality, research-driven product recommendations. According to OpenAI, the model works well in complex categories like electronics, beauty, home & garden, kitchen appliances, and sports & outdoor gear — areas where a traditional shopping prompts in ChatGPT often fall short. And in a strategic move to drive adoption, the new shopping experience is free for all logged-in ChatGPT users during the holiday season. It’s also available in ChatGPT Pulse, currently offered to ChatGPT Pro users. When relevant, Pulse can proactively suggest personalized buyer’s guides based on your past conversations. For example, if you’ve been talking about e-bikes, a future Pulse card could surface recommendations for the best accessories. This reminds me of the Walmart Chile example given in last week's earnings call. ‘Carrito Listo’ is where Walmart Chile uses generative AI to create a customers’ order, send a WhatsApp prompt asking if they want to buy that basket, and all them to buy it. It accounts for 20% of the ecommerce business in Chile now. Another call out from OpenAI is that if memory is turned on, the results reflect the user's past preferences, allowing for more personalized product recommendations. 🛒 Five core use cases OpenAI gave for Shopping Research: • Discover new products • Find lookalikes • Side-by-side comparisons • Gift selection • Deal-finding ⚙️ How it works The user chooses Shopping Research from the (+) menu, enters the prompt, and refine results by marking products as “not interested” or “more like this.” After a few minutes, ChatGPT returns a curated guide with up-to-date details pulled directly from retailers. To complete a purchase, the user has to click on a link to the retailer's website. However, OpenAI said it plans to rollout purchases with Instant Checkout to this feature in the future. 🔐 Privacy, neutrality and accuracy OpenAI reinforced that chats aren’t shared with retailers. The results are not influenced by ads or paid placement. Although this model is more reliable at citing product details according to OpenAI's internal evaluation, it can still miss on price or availability — so users should still verify on the merchant’s site.

Will this be the new way to search on Amazon? 🔍 🤖 Amazon recently rolled out #Rufus, its AI-powered shopping assistant, to all US customers. Rufus can answer customer questions, provide product comparisons, and offer tailored recommendations by leveraging Amazon's extensive product catalog and information from across the web. 👟I took Rufus for a test drive. Check out the video below in which I ask, "Hey Rufus, please give me options for running shoes that are suitable for Seattle weather." It accurately understood that Seattle's rainy climate calls for weatherproof shoes with good traction and provided apt recommendations. I find two other use cases of Rufus particularly useful: 📦 𝗔𝗰𝗰𝗲𝘀𝘀𝗶𝗻𝗴 𝗰𝘂𝗿𝗿𝗲𝗻𝘁 𝗮𝗻𝗱 𝗽𝗮𝘀𝘁 𝗼𝗿𝗱𝗲𝗿𝘀: I no longer have to scroll endlessly to find previous orders. I can just use prompts like "Please find the fan I ordered last year" to pull up my past purchases. ⭐️ 𝗥𝗲𝘃𝗶𝗲𝘄𝗶𝗻𝗴 𝗰𝘂𝘀𝘁𝗼𝗺𝗲𝗿 𝗳𝗲𝗲𝗱𝗯𝗮𝗰𝗸 𝗼𝗻 𝗽𝗿𝗼𝗱𝘂𝗰𝘁𝘀: Instead of reading through numerous comments and reviews, I can simply ask "What are customers' biggest complaints about this product?" to get a concise summary for product I'm interested in. 💡Rufus performs well for specific requests like these, and as it's still in beta, I hope to see it improve for broader and more complex queries. It'll be interesting to see whether Rufus becomes widely adopted or if customers will stick to the traditional search bar and directly reading customer reviews. 💬 What are your thoughts on AI shopping assistants? Are they the future of e-commerce or just another tech trend? #Amazon #RufusAI #DareToDO #ECommerce #AIShoppingAssistant