The web’s newest customers are autonomous AI agents that shop, negotiate, and even replenish inventory on behalf of users. In this business-to-agent (B2A) model, companies build products and rails (infrastructure) for software agents first, humans second.
A recent Medium essay that popularized the term frames agents as “always-on power users” that expect perfect APIs, structured product data, and instant fulfillment decisions.
Why CPGs and Retailers Should Take Notice
Many investment analysts forecast B2A sales increases for big players like Meta, Amazon, Bank of America, and Booking.com. These companies are projected to be some of the early winners in creating agents that can discover content and complete transactions.
Google is already piloting agentic checkout, letting Gemini add items to a cart and pay, sometimes without user confirmation. Fintech firms are lining up too, with Visa, Mastercard, and PayPal all demoing agent-ready APIs at this year’s FinovateSpring conference.
Enterprise AI is also proving its worth. SymphonyAI and others show how supply chain agents can optimize pricing, planograms, and demand forecasts autonomously, boosting revenue and cutting waste. Early pilots point to double-digit efficiency gains, which are far easier to defend than seasonal promotions.
As these capabilities roll across commerce, the addressable market for agent-initiated purchases could dwarf today’s social commerce spend, which eMarketer forecasts will exceed USD100 billion in the US this year.
How B2A Changes the Rules of Engagement
This new paradigm is rewriting the traditional rules of commerce:
| Old rule (B2C) | New rule (B2A) |
| Emotional storytelling wins share of mind. | Structured data — real-time prices, inventory, and ESG scores in machine-readable feeds — wins share of algorithm. |
| The funnel is linear (awareness to conversion). | Loops are instantaneous. Agents iterate options and repurchase as soon as data inputs change. |
| Loyalty = points + brand love | Loyalty = cheapest total cost to serve. Agents are ruthless arbiters unless constrained by user preferences. |
| Trade spend targets human category managers. | Trade spend targets agent APIs. Incentives and digital coupons need to be machine triggered and verifiable. |
Competitive Advantage for First Movers
As agents begin to dominate decision-making and purchasing, brands that adapt their infrastructure and data strategies stand to benefit in these ways:
- Shelf before search: When an agent scans the market, the first brands it “sees” are those with open, standardized product graphs. Early movers effectively pre-empt the shelf.
- Margin defense: If a CPG firm pipes real-time price-pack SKU-level cost data into an agent, the agent can automatically adjust pack sizes or formulations before a price war erupts, preserving gross margin.
- Zero-click replenishment: Retailers with trustworthy inventory APIs that let household agents reorder staples without the consumer ever reopening an app will lock in market share.
- Data flywheel: Serving agents generate granular intent signals (why the agent chose SKU A over SKU B). These can feed back into CPG innovation cycles faster than traditional panel data.
Who Is Building the Rails?
The infrastructure driving the shift to B2A is already under construction.
- Y Combinator is funding internal agent builders and full-stack AI firms that sell straight to agents.
- Google has integrated a Gemini-powered agentic checkout into search and YouTube.
- Amazon is testing Rufus and Alexa agents that research and purchase on Prime.
- Meta’s AI assistant can place ads and transact across WhatsApp and Instagram Shops. Meta could be the big winner since its data often knows users better than other platforms do.
- Visa, Mastercard, and PayPal are creating API suites for autonomous payment authorization.
- AutoCart and other startups are building agent-optimized product data hubs feeding retailers and CPGs.
Getting Started: A CPG and Retailer Checklist
Here’s how brands and retailers can get started in B2A commerce:
- Publish an agent-ready product catalog (with nutritional information, sustainability scores, and live pricing, for example).
- Expose real-time inventory and fulfillment APIs at the DC and store level.
- Run an agent-friendly trade promotion pilot using machine-verifiable discount logic.
- Upskill master agents (agents that lead other agents) and treat them like the mobile wave circa 2010.
- Invest in B2A data and analytics that dig into why agents choose rival SKUs. Fine-tune your product attributes accordingly.
Just as B2C pioneers seized many first-mover advantages in ecommerce, brands that optimize for AI agents today will own the invisible shelf tomorrow. B2A is an imminent distribution channel that could rewrite category shares in 18-24 months. Brands should act now to make these agents their private sales forces, and not their competitors’.
For a robust assessment of big picture trends and more agile scenario planning that can flex to current and future challenges, check out Kantar’s Business Not As Usual framework.
