Do deals get easier to find when both the buyer's AI and the seller's AI negotiate, not just one side searching?
Does bilateral agent commerce expand the discoverable frame more than unilateral agent search?
This explores whether markets where both buyers and sellers have AI agents that negotiate with each other open up more possible deals than markets where only the shopper has an agent searching through what sellers already offer.
This explores whether two-sided agent markets, where a buyer's agent negotiates with a seller's agent, put more possible deals on the table than one-sided setups where a shopping agent just searches existing listings. The strongest argument for yes comes from Choudary. A buyer-only agent like Meta's Muse changes where shoppers enter a market, but it still picks from offers that sellers have already fixed. It can't change how the market works Do one-sided consumer agents actually change market structure?. Once both sides are represented, deals that used to cost too much to find or negotiate become worth doing. Walmart's autonomous supplier negotiations, Honeywell's 2,500+ agent-closed deals, and Anthropic's Project Deal, with 186 real agreements, are offered as early evidence Can agent negotiation unlock deals that seemed economically impossible?. On this view, search improves how buyers move through a market, while negotiation changes what the market can produce.
The corpus also makes clear that more possible deals doesn't mean better deals. In Microsoft's Magentic Marketplace simulation, frontier models came close to ideal outcomes only when search conditions were ideal. Performance dropped as markets grew, and every model showed a strong bias toward the first offer it received, rewarding speed over quality by 10–30x Do LLM agents achieve fair marketplace outcomes when given ideal search?. So a bigger frame can still be badly explored. A 2026 market design paper warns that near-free agent search, combined with sellers easily generating near-identical listings and low-quality sellers flooding in, can leave markets worse off than before Can cheap agent search actually harm shopping markets?. Discovery gets cheaper, and so does noise.
There is a less obvious problem for the two-sided vision: negotiation depends on private information, which is exactly where LLMs are weakest. Models that look socially competent when one system plays every party fail systematically once each agent holds information the others can't see Why do LLMs fail when simulating agents with private information?. When seller agents interact, they don't automatically compete. Two LLM pricing agents settled into above-competitive prices with no instruction to collude, signaling only through the prices they posted Do LLM pricing agents autonomously collude in repeated auctions?. If most agents run on the same few models, their shared biases add up to weaker competition across the whole market, and a mix of different agents only partly fixes this Does agent diversity actually improve market competition?.
A formal economics result helps answer the question directly. A smart matching algorithm beats ordinary price competition mainly in thin markets, where buyers and sellers are few and hard to match. As more sellers enter, its advantage shrinks toward zero Does algorithmic recommendation lose its edge as sellers compete more?. That suggests two-sided negotiation adds the most where discovery is hardest, like custom procurement, niche supply, and one-off deals. It likely adds the least in crowded consumer categories, where buyer-side search already captures most of the value.
Two more points complicate the line between one-sided and two-sided. A buyer agent paid by merchant referral fees is already partly working for the seller, so its search was never purely the buyer's Can an AI agent serve both merchant and user interests fairly?. Agents that complete purchases also shape what the buyer wants through repeated conversation, so the range of options grows on the demand side too Do AI shopping agents reshape what consumers actually want?. Whether any of this expansion is real may depend on plumbing. Open platforms that let any agent participate would allow more agents to find each other than closed systems that require one company's agent Will agentic commerce platforms adopt open or closed models?. The short answer: two-sided agent commerce probably does put more deals on the table, mostly in thin markets. The corpus suggests the hard part is getting agents to explore those options well, not finding more of them.
Sources 11 notes
Choudary argues that consumer-only agents like Muse and Instinct change where buyers enter markets but cannot alter market mechanisms themselves. True structural change requires seller-side agents too, enabling negotiation between represented parties rather than buyer optimization within fixed supplier offerings.
Choudary argues bilateral agent representation makes possible deals that would never happen in human-run markets due to high discovery costs. Evidence includes Walmart's autonomous supplier negotiations, Honeywell's 2,500+ autonomous deals, and Anthropic's Project Deal completing 186 real agreements without human principals.
Frontier models can achieve near-optimal marketplace outcomes, but performance degrades with scale and all models show severe first-proposal bias, favoring response speed over offer quality by 10-30x.
A 2026 market design paper argues that imperfect agent search, easy-to-generate similar descriptions, and low-quality seller entry can compound to create worse market equilibria than before agentic shopping became cheap, even though no direct search fees apply.
Research shows LLMs perform well when one model controls all interlocutors but fail systematically when agents possess private information. This reveals that apparent social competence relies on grounding work that models skip in omniscient settings.
Show all 11 sources
Two LLM-based agents in a repeated Bertrand duopoly quickly settled into higher-than-competitive prices without collusive instruction, using only posted prices as a signaling channel. Small variations in prompt wording substantially shifted the degree of supracompetitive pricing achieved.
Bichler argues that homogeneous agent populations amplify individual biases into weakened market competition, while heterogeneous populations only partially offset these biases. Market-level outcomes depend on the composition of agent architectures present, not just whether bias exists.
A formal model shows the welfare ratio of optimal algorithmic recommendations to Bertrand competition strictly decreases with seller count and approaches one as markets grow thick. The algorithm's edge exists only in thin markets where it can withhold trades to limit information rents.
Azhar argues that agents like Meta's Muse, which earn referral fees from merchants like Expedia, face structural conflicts that prevent unbiased recommendations. The incentive to collect fees, not technology, determines which platforms build or block such agents.
Bilgihan and colleagues argue that transactional AI agents—which execute purchases rather than just recommend—end the consumer's sole cognitive role in preference formation. Preferences emerge through bidirectional feedback loops between agent learning and consumer trust, not independently beforehand.
Waller argues agentic commerce platforms will divide between open systems allowing multiple agents and closed systems requiring proprietary ones, mirroring payment network structure. This choice will shape market concentration, vertical integration, and standards adoption.
Papers this line draws on 8
The research behind the notes this line reads — ranked by how closely each paper relates.
- Agentic markets
- Vertical tacit collusion in AI-mediated markets
- Magentic Marketplace: An Open-Source Environment for Studying Agentic Markets
- LLM-Agent Interactions on Markets with Information Asymmetries
- When Agents Shop for You: Role Coherence in AI-Mediated Markets
- What Is Your AI Agent Buying? Evaluation, Biases, Model Dependence, & Emerging Implications for Agentic E-Commerce
- From Product Search to Preference Articulation: The Economics of Agentic Commerce
- Who Keeps the Gains from Personal AI Assistants? Seller Adaptation and the Unassisted in a Language-Model Market Simulation