AI makes it possible to build faster than ever. However, none of that speed applies to the data. Lose the data that users give you, and you have to start all over again. Let agents iterate over the code, but leave the data to humans.
Walmart is using limited-time draws to sell scarce collectibles, including basketball cards, NeeDoh toys, and Pokémon boxes. Shoppers enter during a set window and are randomly selected instead of racing to checkout. The format creates anticipation and gives collectors a reason to discuss entry rules and results online. It shows how intentional friction can turn scarce products into experiences rather than simple transactions.
B2B buyers increasingly use creators for both early research and later-stage validation. However, 80% of B2B marketers say finding credible creators is their biggest hurdle. LinkedIn recommends using Creator Marketplace to vet experts by industry, role, and audience, then amplifying their content with Thought Leader Ads or BrandLink. Thought Leader Ads deliver a 252% higher CTR than standard image ads, while BrandLink campaigns see 130% higher video completion rates than in-feed ads.
Jev is a new decision model that returns labels, scores, and yes/no answers with calibrated confidence instead of generated text, making it a cheap fit for classification and routing in Clay. It connects through an HTTP API column in about 10 minutes. Keep the input state short, since extra context can lower accuracy. Add an "unclear" option and write criteria that don't overlap. Set stricter confidence thresholds for hard-to-undo actions, then gate paid enrichment and LLM writing on the result.
AI shopping assistants could weaken marketplaces' advertising and commission economics by owning the customer relationship and sending orders directly to fulfillment providers. Amazon generated $69 billion in advertising revenue in 2025, so losing control of product discovery could threaten a major profit pool even if it keeps fulfilling orders. Platforms like Shopify and Toast have more reason to welcome agents because they can gain new sales while still earning software and payments revenue. The outcome depends on whether agents create new demand or simply reroute existing orders, since new purchases can offset lower profit per order.
Strong PMMs are struggling to land jobs because companies are hiring cautiously and looking for candidates who fit specific business needs. Job seekers experience three common roadblocks: broad positioning, forgettable interviews, and explaining experience instead of demonstrating how you would solve the company's problems. The author recommends targeting your positioning around a specific type of company and using a consultative interview approach that diagnoses the company's needs, then structures answers around Process, Situation, Action, and Result. You can also stand out by creating concrete proof of how you would solve a company's problems, such as one candidate who built an ROI calculator for a final interview and had it shared across the company.