Shopping Agents Have No Mercy. Sellers Need a Way In.
AI shopping agents will compare every offer in one tap, so sellers of identical goods will win on how easily an agent can order from them.
Phia, the AI shopping app co-founded by Phoebe Gates and Sophia Kianni, raised a $35 million Series A at a valuation of about $185 million in January 2026. Its “Should I Buy This?” button compares any item with real-time retail and secondhand prices and tells the shopper whether the price is fair. The idea resembles Google Flights. For sellers, that comparison is the part worth paying attention to.
Key ideas
- 1. Price stops protecting anyone who sells identical goods. When an agent sees every offer at once, the cheapest one wins unless something real makes the product different. Points programs don't change the math: an agent converts them into dollars and keeps comparing. Airlines learned this lesson the hard way, and their answer was loyalty, which grew so valuable that United valued its MileagePlus program at about $22 billion in 2020. Agents are far less sentimental than frequent flyers.
- 2. Discovery moves to agents. Fulfillment stays with whoever can deliver. Finding the best offer is easy for an agent. Ordering, paying and delivering is still hard, so agents lean on marketplaces as an order desk. The legal ground is shifting: in August 2026, a US appeals court overturned a preliminary injunction in the Amazon–Perplexity case, finding that the user directing the Comet assistant was the one accessing Amazon. The case continues, but stores may not get to keep agents out simply by treating them as unauthorized third parties.
- 3. The pipe is still taking shape. Standards are emerging: Google's UCP is designed to work with existing retail infrastructure and AP2, A2A and MCP. In March, OpenAI pulled back from its original in-chat checkout rollout and shifted toward purchases on merchants' sites. New routes continue to emerge: the Safeway experience in ChatGPT now carries grocery shopping through checkout. The sellers who become easy for agents to buy from, on every channel, will set the pace.
What to do now
Ask ChatGPT, Gemini and Perplexity to find and buy your product. See whether they find you, understand what makes you different, and know when it ships. Then decide which game you're playing: price and operations, or a brand worth asking for by name. Both need clean product data, honest stock and a delivery promise a machine can trust.
WIZIUM perspective
Where WIZIUM stands
WIZIUM is building an operating environment on the seller's side, where people and AI agents work on the same picture of the business. The goal is to connect a seller's products to every shelf where buyers and their agents look, from marketplaces to the seller's own store to AI assistants, and keep track of what each channel needs and where the product performs best. This is our assessment of where agentic commerce is heading, not a description of what every channel supports today.
Research Lab
WIZIUM Research Lab follows how commerce is changing and talks to the people living through it. If you sell, build or run operations for brands, your view of agent-driven shopping is welcome. The future of commerce is something we build together.