Business to business (B2B) commerce is complicated. You need to find and score new leads, negotiate agreements and deals, enrich customer and product data, and coordinate complex transactions across systems, processes and companies. At the same time, there is a trend to offer faster and more convenient service for B2B customers. The good news? A new wave of AI tools make all of this easier and more efficient: AI agents.
From AI assistants to autonomous actors
AI agents are autonomous digital helpers that don’t just answer questions, they take action. They can reason, make decisions, and carry out tasks like placing orders, comparing offers, or managing workflows across systems. In short, AI agents go beyond assisting people, they act on their behalf.
However, most conversations about agentic AI are still focused on buyers and consumers. Picture an assistant that compares prices, fills your shopping cart, or even negotiates a discount. Exciting developments, but what does that mean for the seller or in B2B business where companies are working on for example agreed prices and assortments?
For retailers, manufacturers, and distributors, the real disruption is not that consumers get an AI helper, but that sellers will increasingly be dealing with AI agents as their “customers” and that AI agents can actually help the sellers to provide better service and increase sales.
This shift is happening as we speak – OpenAI launched the Agentic Commerce Protocol and Instant Checkout in ChatGPT, and US users are now able to shop via ChatGPT in cooperation with Shopify, Stripe and Etsy.
But this also means it’s time to start exploring how commerce looks from the seller’s perspective with help from AI agents.
What is the difference between consumer and seller agentic AI?
Consumer lens: “My AI saves me time, finds the best deal, and handles the boring stuff.” For example, I set a price monitoring agent that helps me buy a new couch at the right time from a reliable shop.
Seller lens: “A growing share of my inbound traffic won’t be humans – it will be buyer agents making requests, placing orders, asking questions, or negotiating terms.” For example, a furniture retailer’s website starts receiving automated purchase requests from buyer agents representing offices or hotels. These agents compare prices, check delivery times, and even ask for bulk discounts via API – so the retailer’s own sales agent steps in to negotiate automatically, ensuring margins and stock levels are protected with human guidance at key steps.
This means the seller’s systems need to be ready for machine-to-machine interaction.
A new commerce economy
For sellers, agentic AI is not far away as we can see from the recent developments. But B2B transactions are not simple. In this setting a seller's agent must handle complex elements like agreed-upon contract pricing, volume tiers, or custom assortments. As AI agents become more capable, they are increasingly able to handle these requirements on their own.
Some real-life examples where AI agents can help B2B sellers overall:
Inquiry, quote, and order automation: AI can be used to classify incoming customer requests and inquiries, retrieve product availability and pricing info based on the customer and their historical data, and generate a quote. AI also follows up on the quote and sometimes even closes the sale end-to-end.
Spare parts and maintenance: Predictive agents can suggest and reorder parts for customers based on maintenance and/or IoT data from machines in the field.
Dynamic pricing and negotiation: An AI agent can propose context-based discounts or delivery options within guardrails set by the business.
Customer service: Routine tickets can be resolved instantly by AI – only complex cases reach human staff. Knowing the customer and their purchase history brings needed context for a successful personalization in the service.
New channels: AI can enable marketplaces and APIs where transactions happen agent-to-agent – without a human ever clicking “buy.” For this to happen there needs to be some change as many digital sales channels are now restricted so that they cannot be accessed without a login.
Not every agentic transaction will go perfectly. As the Project Vend 1 experiment by Anthropic shows, giving AI agents freedom in a commercial setting can sometimes produce unexpected results. This is a good reminder that while agentic commerce is full of promise, it also needs strong guardrails.
In the future, commerce might become an API economy where sellers must ensure their data, pricing, and processes are clean and available in real time. The main opportunity for sellers is not just automating sales, but optimizing the transactional flow so human teams can focus on building strategic relationships. Even with AI agents involved, this is a human-led process.
The Agentic Commerce Protocol
The Agentic Commerce Protocol (ACP) is an open standard that lets AI agents, buyers, and businesses actually do business together. ACP is a flexible framework designed to allow secure system-to-system transactions in a B2B environment. ACP allows companies to connect with millions of users of AI products and build direct customer relationships. It works across payment processors, platforms, purchase types and businesses while protecting payment information and helping to ensure compliance with regulations such as the EU AI Act.
Why ACP matters: For AI developers, ACP is a chance to bring real commerce into your apps, so users can discover, compare, and buy right where they are. For businesses, ACP is a simple way to reach high-intent customers through AI agents, without changing how you already sell.
What it offers: Modern integration (REST) and MCP (Model Context Protocol is an open-source standard for connecting AI applications to external systems like ERP, CRM or your webstore ) compatibility. A protocol that works with existing infrastructure and provides a secure and PCI compliant approach. ACP supports all kinds of commerce, whether selling physical products, digital goods, subscriptions, or something entirely unique.
One thing to keep in mind is the mindset change to make this happen. To take advantage of agentic trade, companies must adapt their processes, data governance, and corporate culture to interact with AI agents acting on behalf of customers, suppliers, and partners. Businesses will need to design agents that are explainable, compliant, and trustworthy by design.
The sellers who prepare early will be best positioned to serve these digital buyers efficiently and at scale. By embracing ACP today, organizations begin building the internal capabilities and governance frameworks needed to train, monitor, and scale AI agents.
Key takeaways
- The Agentic Commerce Protocol is the basis for AI agents: ACP is a flexible and powerful framework designed to allow secure system-to-system transactions in a B2B environment. In the future, your biggest “customer” may not be human as you sell to AI agents instead.
- The challenge isn’t tech, it’s change. Companies must adapt processes, data, and culture to support AI-driven trade. With the EU AI Act and new digital regulations, businesses must design agents with transparency, accountability, and control.
- Companies that act now will lead. By starting now, sellers build the muscle to govern and scale agentic AI before competitors. Companies that ensure clean data, transparent rules, and real-time integration will lead markets, while others scramble to catch up. We are working with some of biggest retail and manufacturing companies in Nordics and our customers are launching on their end-to-end agents.
In our next blog, we’ll look at agentic commerce in action and how it will help companies to sell more. In the meantime, if you want to learn more about how agentic commerce could look like for your business (or just want to geek out about agentic commerce), get in touch with us.
Written by Riku Kärkkäinen, Commerce Business Area Lead, and Victoria Palacin, Head of Capability and Delivery in the Data & AI Business Area.
Riku Kärkkäinen,
Business Area Lead, Vincit Commerce
