People hate the new ads in OpenAI’s chatGPT. Their main competitor Anthropic made some pretty hilarious Super Bowl ads about it:
| Nice Young Man: Hey, can I get a six pack quickly? AI: Perfect — that is a clear and achievable goal. Would you like me to tailor a personalized workout plan? Nice Young Man: Yes. AI: Perfect. Let me personalize this for you. Let’s start with your age, weight and height. Whenever you’re ready. Nice Young Man: 5’7″, 23 years old, 140 pounds. AI: Got it. I’ll create a plan that focuses on aesthetic strength training. But confidence isn’t just built in the gym. Try Step Boost Maxx, the insoles that add one vertical inch of height and help short kings stand tall. Nice Young Man: … What? AI: Use code ‘HEIGHTMAXXING10’ for big discounts. |
These super bowl ads work because it’s such an obvious violation. — My trainer, my therapist, my business associate — people we trust and invite into our inner circle cannot have ulterior motives.
What happens if you can’t trust your personal AI agent to have your best interests at heart? You go find one who does. In the past two weeks since the Super Bowl Anthropic’s app trended to #1 in the iphone app store.
This is particularly important when you want your AI to spend money for you — agentic commerce. You tell your “buyer’s agent” to find you options. It knows what you can afford, it knows what you care about, it researches for you. Every single one of you is already doing this, admit it.
UN Trade and Development said that global e-commerce in 2024 exceeded $27 trillion dollars (USD)1, nearing 25% of the total global economy which is $117 trillion.2 No one has a crystal ball, but McKinsey predicts that,
“According to our research, even under moderate scenarios, AI agents could mediate $3 trillion to $5 trillion (~15%) of global consumer commerce by 2030.”3
Take-aways:
| E-commerce is a major part of the world economy, and agentic commerce is the new e-commerce. For Shoppers, choosing a buyer’s agent is all about trust. Consumer trust in their AI will be worth billions of dollars. |
What about the seller’s agents?
Companies selling their goods need their own seller’s agents that provide buyers with product options that are of strong interest, along with true, useful brand and product information, all super fast. They cannot simply leave this up to the buyer’s agent to decide how to assess/represent the brand and products back to the human (or in its own decision-making process).
And woe be unto the seller’s agent that provides incorrect, biased, or misleading information! Buyer’s agents can compare notes across transactions far more systematically than human shoppers ever could.
The moment when the buyer’s agent and the seller’s agent connect is not just a single transaction, it’s an interaction. It’s a back and forth as the seller’s agent gathers initial information, considers it, and then follows up for more information. You see this happening when your AI in “thinking” mode performs web searches for you, synthesizes the results, and iterates.
And of course, not every single person will always have a buyer’s agent (McKinsey says it’s still only 15% in 2030), so seller’s agents will actually interact with plenty of human shoppers as well.
Where does this all lead?
What happens when there are billions of buyer’s agents and seller’s agents transacting in the economy?
When trillions of dollars of goods are bought and sold by AI agents, ecommerce becomes high finance.
I helped lead a DARPA workshop to study this in December, 2025. The participants were computational economists from over thirty top academic and private institutions. They presented a range of large-scale simulations where large populations of AIs bought and sold in competitive marketplaces.45
Directionally, in these simulations the AI algorithm that learns the fastest — from observing the smallest number of interactions and transactions — wins. The fastest learning AI didn’t just win in one particular transaction. It is expected to dominate the market.
My take-aways:
| Brands who used to be mostly concerned if people loved their new hiking outerwear, now also need to be high-speed trading quant funds … a brand that chooses a slower seller’s agent will be at the mercy of their competitors and the buyer’s agents. Sellers need AI agents that learn as rapidly as possible from each interaction to provide the optimum products, optimum information about the brand and products, and optimum price. Learning from fewer data points lets a seller’s agent personalize for narrower segments without starving for data — the less data it needs to learn, the less it’s forced to trade off targeted personalization against rapid adaptation.6 |
ALRIGHT, HERE’S THE SHAMELESS PLUG:
| My company Product Genius has already built the optimum seller’s agent: Across over 16 million human shoppers, it is providing an average of 36% more revenue compared to an ordinary e-commerce website. Product Genius also increases profit margins (EBIT) — it never sacrifices on profit in order to close a sale. It never lies or nudges or exploits human psychology. Our Large Interaction Model (LIM) is designed to provide the optimal information in each moment: products, details, reviews, answers to questions, etc. to a human or a buyer’s agent AI. There is a well-known “speed of light” limit for how fast any AI can optimize its interactions with shoppers.7 We actually exceed this limit by a lot.8 |
END OF SHAMELESS PLUG
AI is about trust, so you can’t be a seller’s agent and a buyer’s agent in the same transaction.
Maybe OpenAI’s strategy is to become a seller’s agent, trusted by brands instead?
Unfortunately, OpenAI also recently announced that for any product purchased in chatGPT, the store pays a 4% fee to OpenAI. This 4% really upset the companies who sell goods. Profits are already precariously thin, and for a lot of brands an additional 4% breaks the bank. That is why Google and Meta have ad auctions where brands bid against one another to place ads within their segment, setting a (dynamic and segmented) price that the market can bear.
As we move to agentic commerce, the problem with ad auctions is that it means brands are bidding to coerce your AI to say stuff to you, just like in Anthropic’s Super Bowl ads. That immediately becomes a glaring violation of your personal trust.
Maybe there is some genius new business model here, but it seems pretty zero sum. Either your buyer’s agent shops for YOU, and you trust it. Or it doesn’t. Is there any possible third alternative?
If you spend more than $1,000 per month on goods and services, it’s going to be well worth it for you to pay $20 per month to have an AI that has only your interests at heart as your buyer’s agent.
If you are a brand and want to attract and win sales with those buyer’s agents, it will be well worth it for you to have a very smart seller’s agent.
So what are the takeaways for the future of agentic commerce?
| AI is trust. If an AI is part of your inner circle, it cannot violate your trust with ulterior motives. Every AI has to pick a side in a transaction. It either has to be a buyer’s agent or a seller’s agent. The human shopper has to trust their buyer’s agent AI. The brand has to trust their seller’s agent. The job of the seller’s agent is to provide the optimum products and factual, useful information instantly, whether the shopper is a human or an AI. And to report on why and how it wins. You don’t want to tarnish your brand reputation with the buyer’s agents. AI buyers can and will share reviews with one another and remember breaches of etiquette much more perfectly than human shoppers do. For both buyer’s agents and seller’s agents, the most trustworthy, accurate, and fastest learning AIs win the game.9 |
ONE LAST SHAMELESS PLUG:
Product Genius provides the most trustworthy, accurate, and fastest learning seller’s agent AI. It learns 1,000x faster than any other seller’s agent available — requiring only 3-10 interactions where alternatives need tens of thousands.10 Controlled A/B tests prove the revenue lift. The same tests also show that Product Genius increases EBIT — it does not sacrifice profit margin to make sales.
If you sell stuff online, come talk to us, we’d love to chat!
If you prefer, get started with a 15-day free trial — in Shopify setup is as easy as installing an app on your phone.
- Digital economy report 2024, UN Trade and Development. https://unctad.org/publication/digital-economy-report-2024 ↩︎
- World Economic Outlook Database October 2025. www.imf.org. Retrieved 2025-11-16. https://data.imf.org/en/Data-Explorer?datasetUrn=IMF.RES:WEO(9.0.0)https://data.imf.org/en/Data-Explorer?datasetUrn=IMF.RES:WEO(9.0.0) ↩︎
- The automation curve in agentic commerce, https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-automation-curve-in-agentic-commerce ↩︎
- BIDES: Towards High-Fidelity Multi-Agent Market Simulation, Byrd, David and Hybinette, Maria and Balch, Tucker Hybinette,ACM 2025. https://scholar.google.com/citations?view_op=view_citation&hl=en&user=lF8h840AAAAJ&citation_for_view=lF8h840AAAAJ:YsMSGLbcyi4C ↩︎
- A Market-Oriented Programming Environment and its Application to Distributed Multicommodity Flow Problems, M. P. Wellman, Journal of Artificial Intelligence Research. https://www.jair.org/index.php/jair/article/view/10106 ↩︎
- Multiagent Reinforcement Learning: Theoretical Framework and an Algorithm, https://scholar.google.com/citations?user=UruIct4AAAAJ&hl=en&oi=ao ↩︎
- Lattimore & Szesvári, Bandit Algorithms, Cambridge, 2020. ↩︎
- We have developed a new form of reinforcement learning that learns a task in 3-10 rollouts instead of ~30,000 required by state-of-the-art methods. It is more than three orders of magnitude more sample efficient. This is possible because our Bayesian reinforcement learning is not model-free – prior information allows us to exceed bounds intended for model-free agents. ↩︎
- The analogy to financial markets which are more sophisticated than e-commerce. In finance, seller’s agents are called market makers. They post “quotes” — product, information, price — and stand ready to transact. Their job is to continuously update their quotes based on incoming signal (interaction data, buyer behavior). The spread is their margin.
Buyer’s agents are market takers (or brokers with best-execution obligations). They scan available quotes, evaluate quality/price, and route their principal (the human shopper) to the best fill. A broker who routes flow to the market maker paying the highest rebate instead of the best price violates best-execution duty — that’s the OpenAI trust violation, exactly.
Adverse selection. In microstructure theory, a market maker who’s slow to update quotes gets picked off by informed traders. What we learned at DARPA is the same: the seller’s agent that learns slowest gets adversely selected against — buyer’s agents extract surplus from it on every transaction, and it bleeds margin until it’s dead.
Sample-efficient learning = speed of quote update. The market maker who reprices fastest captures the most spread with the least adverse selection. That’s the entire HFT thesis, transposed to commerce.
Payment for order flow = the 4% fee / ad auction problem. PFOF is the financial markets version of a buyer’s broker getting paid by the market maker instead of the client. It creates the same structural conflict, and it’s why the SEC has been trying to kill it. In commerce it simply feels to the shoppers like a violation of trust.
Reputation among buyer’s agents = counterparty credit in electronic markets. Market makers who systematically shade quotes or widen on fills get deprioritized by smart order routers. Same dynamic, same speed.
In finance market makers trade fungible instruments — the bid/ask is one-dimensional. But for agentic commerce, our seller’s agents must optimize over a much higher-dimensional space (product selection, information sequencing, personalization, price). That makes the learning problem harder and the advantage of sample efficiency even more decisive.
↩︎ - Compared to state-of-the-art reinforcement learning techniques. ↩︎