AI-native commerce: agents that shop, sell, and negotiate

1) Market thesis

AI-native commerce is moving from human-led search, comparison, and checkout to machine-led intent execution, where agents discover products, represent buyers and sellers, negotiate terms, and complete transactions across increasingly programmable commerce rails.

2) Market map

Segment 1: Consumer shopping and discovery agents

Companies:

• Amazon — Rufus
• Google — AI Overviews / Gemini shopping experiences
• Perplexity — Shopping
• Klarna — AI shopping assistant
• Daydream — AI-native fashion shopping platform
• Microsoft — Copilot shopping features

Dynamics: These products compete for the consumer’s commercial intent before it reaches a retailer’s website; differentiation depends on product data quality, recommendations, trust, transaction completion, and access to broad merchant inventory. Search incumbents have distribution, while Amazon and Klarna have stronger transaction and purchase-history data.

---

Segment 2: Merchant-side selling, service, and commerce-operations agents

Companies:

• Shopify — Sidekick and Shopify Magic
• Salesforce — Agentforce Commerce
• Gorgias — AI customer-service and ecommerce agents
• Bloomreach — Loomi AI
• Adobe — Commerce AI capabilities
• Intercom — Fin AI Agent

Dynamics: These agents automate merchandising, customer service, marketing, retention, and parts of sales operations. The main battleground is the merchant system of record: platforms that own catalog, customer, order, and workflow data can move from copilots to agents that actually execute changes and transactions.

---

Segment 3: B2B procurement and negotiation agents

Companies:

• Pactum — autonomous supplier negotiations
• Fairmarkit — autonomous sourcing and tail-spend optimization
• Keelvar — sourcing orchestration and optimization
• Arkestro — predictive procurement and supplier bidding
• Zip — procurement intake and workflow automation
• Icertis — contract lifecycle management with AI capabilities

Dynamics: B2B negotiation is attractive because contracts are large, repetitive, and economically measurable. Adoption is slower than in consumer commerce because agents need authority limits, compliance controls, supplier participation, and integration with ERP and procurement systems.

---

Segment 4: Agentic transaction, identity, and payment rails

Companies:

• Visa — Intelligent Commerce
• Mastercard — Agent Pay
• Stripe — agentic commerce capabilities and payment infrastructure
• PayPal — agentic commerce initiatives
• Shopify — Agentic Commerce Protocol
• Worldpay — AI and account-to-account commerce infrastructure

Dynamics: This layer is becoming the control point for authentication, delegated purchasing authority, payment credentials, fraud liability, refunds, and settlement. Standards are still unsettled: wallets, card networks, commerce platforms, and model providers all want to define how an agent proves who it represents and what it is allowed to buy.

Note: Product names and launch scopes in this category are evolving quickly; the companies are real, but some offerings remain early-stage or geographically limited.

---

Segment 5: Commerce data, product catalogs, and agent infrastructure

Companies:

• Salsify — product experience and catalog data
• Akeneo — product information management
• Reltio — customer and product data unification
• Algolia — search and discovery infrastructure
• Constructor — ecommerce search and recommendations
• Coveo — AI search and relevance infrastructure

Dynamics: Agents are only as reliable as the product, price, inventory, policy, and merchant data they can access. This segment supplies the structured data, retrieval, ranking, and APIs needed for agents to make accurate recommendations, although much of the value may be captured by larger commerce platforms that bundle these functions.

3) Which segment captures the most value?

Likely winner: agentic transaction and commerce-control infrastructure

The transaction and control layer is best positioned to capture disproportionate value because it can sit at the moment of authorization, payment, fulfillment, fraud prevention, and settlement. It can monetize every transaction rather than only software seats or advertising impressions.

Its advantages include:

1. Per-transaction economics: payment fees, authorization fees, fraud services, and potentially agent-routing fees.
2. Trust and permissions: buyers will need controls over budgets, categories, merchants, refunds, and recurring purchases.
3. Network effects: more buyers attract merchants; more merchants improve agent utility; more transaction data improves risk and recommendations.
4. High switching costs: once identity, payment credentials, order history, and fraud models are embedded, replacement becomes difficult.
5. Strategic scarcity: many agents can generate recommendations, but relatively few institutions can safely authorize and settle purchases.

The major caveat is that consumer platforms may capture the highest gross profit pool if they own the user relationship and can steer demand toward their own marketplaces or advertising products. In practice, the winners may be a small number of vertically integrated platforms—such as Amazon, Shopify, Google, Visa, Mastercard, and PayPal—rather than an independent infrastructure vendor.

4) White-space opportunities

Opportunity 1: A neutral, cross-platform “agent mandate and liability” layer

There is no broadly adopted independent standard or network that lets a consumer specify:

• what an agent may buy;
• maximum price and total budget;
• preferred merchants and excluded merchants;
• acceptable substitutions;
• cancellation and refund rights;
• when human approval is required;
• who bears liability if the agent makes a bad purchase.

Visa, Mastercard, PayPal, and commerce platforms are addressing pieces of this, but they are primarily optimizing their own rails. A neutral authorization layer could provide portable permissions, audit trails, agent identity, dispute handling, and insurance across retailers, marketplaces, banks, and payment networks.

Opportunity 2: A genuinely autonomous cross-merchant B2B buying agent

Current procurement tools generally focus on sourcing events, spend analysis, intake, or supplier negotiation. Consumer shopping agents generally focus on discovery and checkout. The gap is a trusted system that can manage an entire recurring B2B purchase:

1. understand a company’s requirements and policies;
2. compare approved and unapproved suppliers;
3. negotiate price, lead time, service levels, and payment terms;
4. verify certifications and availability;
5. place the order;
6. reconcile delivery, invoice, and contract compliance; and
7. switch suppliers when performance deteriorates.

Adjacent vendors such as Pactum, Fairmarkit, Keelvar, Zip, and Icertis cover important components, but no obvious incumbent owns the complete, cross-system buying loop for mid-market companies. The opportunity is particularly strong in fragmented categories such as industrial supplies, facilities management, packaging, healthcare supplies, and hospitality.