From search to AI-directed shopping
AI recommendation engines in commerce are software systems that continuously analyze shopper behavior, product data, and contextual signals to predict what each person is most likely to want next, and then act on those predictions in real time across livestreams, marketplaces, and ad channels to reduce manual merchandising and campaign work while increasing relevance. This shift matters because it quietly moves power from search boxes and human media buyers to algorithms that decide which products appear on screen, which ads run where, and which messages follow you from a live auction to a conversational AI chat. The platforms making the boldest moves now are the ones that accept this reality and build for it, rather than treating AI as a sidecar tool for small optimizations.
Whatnot bets on AI to keep live commerce from collapsing under its own weight
Livestream shopping only works if discovery keeps pace with the chaos on screen. That is why Whatnot buying Shaped is less a tech upgrade and more a survival strategy for live commerce personalization. When inventory changes every minute and bids spike on unexpected items, yesterday’s recommendation model is useless. Whatnot says it has spent six years cutting recommendation latency from about a day to minutes and now wants this closer to real time. In practice, that means an AI recommendation engine that can watch 500,000 hours of live video and millions of weekly interactions and adjust suggestions mid-stream. The payoff for shoppers is clear: fewer irrelevant lots, more timely nudges on items they might actually bid on, and a feed that feels curated rather than random.
The acquisition also signals that live commerce platforms will need in-house AI brains, not bolt-on vendors. Shaped specialized in real-time recommendation and search, mixing customer data with large language models for personalized discovery. Bringing founder Tullie Murrell and nearly a dozen engineers inside Whatnot to lead a new Applied AI Research group is a bet that live commerce personalization is now a core competency, not a feature request. And it comes after heavy growth pressure: the company recently passed 1 billion orders and raised USD 225 million (approx. RM1,035 million) in Series F funding at a valuation above USD 11 billion (approx. RM50,600 million). At this scale, every extra second of latency or irrelevant suggestion is wasted money on content and customer acquisition.
Pattern turns ChatGPT into another performance channel, not a novelty
While Whatnot attacks the problem inside a single live platform, Pattern is reshaping how brands treat conversational AI: as one more channel in their ChatGPT campaign management stack, not a quirky side project. By adding advertising in ChatGPT into the same system that already runs campaigns on Google, Meta, Snap, TikTok, and ecommerce marketplaces, Pattern gives brands one interface for cross-channel strategy instead of another isolated dashboard. This is ecommerce AI integration in practice, not press-release theory. The platform already draws on more than 77 trillion ecommerce data points about products, categories, and shopper behavior. Now those signals can inform where and how brands show up when someone is using an AI assistant for product discovery, right alongside search and social impressions.
What makes this move more than a simple placement extension is how much control remains programmatic. OpenAI’s ad system lets advertisers manage budgets, bids, creative assets, product feeds, and conversion settings via Ads Manager or an API. Pattern plugs those controls into its existing workflows, then adds automated campaign execution using AI agents that can carry out selected actions across the platform. OpenAI, not Pattern, still decides which ad appears by weighing conversation context, advertiser input, and its auction systems. Ads appear separately from responses and do not shape the answers. The strategic point: brands that hesitate will find their competitors occupying conversational surfaces by default, with performance data flowing into a unified view long before late adopters even test their first prompt.
Automated ad creation turns catalogs into conversations
If the AI recommendation engine is the brain, automated ad creation is the voice. OpenAI lets advertisers connect merchant catalogs with product titles, descriptions, prices, availability, images, and URLs, then automatically create ads for eligible items without a separate setup for each product. Pattern steps in by preparing those catalogs and managing feeds, keeping data fresh so prices and availability match what appears in the ad. This is the real power shift in ChatGPT campaign management: once the system understands the catalog, it can adapt creative on the fly to different queries and contexts. Product-level reporting via OpenAI’s API then lets brands see impressions, clicks, and attributed conversions for each item, down to country and device segments. The manual campaign spreadsheet starts to look like a relic.
Pattern is also layering in AI agents that can execute campaign actions on top of this automated creative engine. Combined with cross-channel conversion data, that means the same system can see which items perform best in search, social, marketplaces, and ChatGPT, then adjust budgets and bids across all of them. This is where multi-channel campaign management stops being about centralized reporting and becomes about centralized decision-making. Creative teams will still shape brand narratives and guardrails, but the day-to-day work of writing, trafficking, and tweaking ads for every SKU is being handed to generative AI. The brands that embrace this will be able to scale personalized messaging at a pace that manual teams cannot match, even if they wanted to.
The new playbook: design for AI-first commerce, not AI add-ons
The lesson from both Whatnot and Pattern is blunt: the future of ecommerce AI integration will not be won by whoever has the flashiest demo, but by whoever rebuilds their core systems around AI decision-making. Whatnot is treating its AI recommendation engine as the operating system for live commerce personalization, pulling expertise in-house and driving toward near-real-time responses in a chaotic auction environment. Pattern is treating ChatGPT as one more native surface in a cross-channel ad machine, with automated ad creation and AI agents stitched into a single performance fabric.
For brands, the implication is clear. Stop asking how to “add some AI” to existing campaigns and start asking what it means when algorithms choose what customers see, in which order, on every surface that matters. Product data quality, clean conversion tracking, and clear guardrails for messaging become strategic assets. As these systems mature, the distance between a live show, a search ad, and a conversational AI suggestion will shrink into a single, AI-driven journey. The companies planning for that world now will dictate the defaults everyone else has to live with later.
