The integration of agentic artificial intelligence (AI) into e-commerce operations presents a significant, yet uneven, landscape for businesses, with Business-to-Business (B2B) companies currently trailing behind their Business-to-Consumer (B2C) counterparts in adopting fully autonomous AI capabilities. This disparity was highlighted by Paul do Forno, global commerce practice lead at Deloitte, during discussions that underscored the nascent stage of agentic AI in commercial transactions, particularly within the complex B2B sector.
Deloitte’s recent findings, based on a workshop conducted in late 2025, revealed that fewer than 24% of participating suppliers had implemented agentic AI within their selling processes. Do Forno articulated a spectrum of agentic AI maturity, ranging from basic functionalities to complete autonomy. His observations suggest that while agentic AI is poised to revolutionize e-commerce, the realization of its full potential, especially in autonomous operations, remains a distant prospect for B2B enterprises. "Agentic AI in e-commerce is still a long ways away in B2B for autonomous, let alone B2C, which will get there much quicker," do Forno stated. He further elaborated on the inherent differences between the two sectors, noting, "In B2B, there’s a bifurcated world. B2B vs. B2C, and B2B always have been behind."
The current application of agentic AI within businesses, as observed by Deloitte, is primarily focused on addressing discrete challenges as part of broader digital transformation initiatives. Instead of a singular AI solution attempting to manage all aspects of a business process, successful implementations target specific "friction points" across various channels and operational areas. This strategic approach allows companies to gradually introduce and refine AI capabilities, building a foundation for more sophisticated applications in the future.
Addressing B2B Pain Points with AI Agents
For B2B companies, agentic AI is beginning to offer tangible solutions to persistent operational bottlenecks. One prominent use case involves streamlining the reordering process. Traditionally, a buyer might need to manually check product availability, delivery timelines, and potential alternatives. Agentic AI can automate these tasks, allowing a shopper to specify requirements, such as needing a product within a week. The AI agent can then autonomously navigate disparate inventory and logistics systems, assess availability, and propose suitable alternatives if the original request cannot be met. "An agent can go off and go look about the different systems, come back and [say]: sorry, that’s not available, but this is available. It’s an alternative but fits your needs," do Forno explained, emphasizing that "the availability to promise, that’s agents that we’re working [on] already."
Another critical area where agentic AI is proving invaluable in B2B is the conversion of traditional procurement documents into actionable orders. B2B transactions frequently involve manual processing of purchase orders (POs), emails, and PDF documents. Agentic AI can ingest these documents, interpret the order details, and seamlessly integrate them into the company’s e-commerce platform. Do Forno identified this as a primary entry point for B2B adoption: "’I’ve got my PO. It’s attached as a PDF. Can you convert this into an order?’" This capability significantly reduces manual data entry, minimizes errors, and accelerates the order fulfillment cycle.
The successful integration of agentic commerce into a B2B company’s technology stack necessitates a robust foundational commerce platform or cloud infrastructure. This core system provides the essential capabilities for managing product catalogs, customer data, and transaction processing. From this base, companies can then extend their agentic AI applications to various channels, including online marketplaces, punchout catalogs, and other integrated systems. This phased approach, characterized by building upon a solid core, allows for incremental deployment and scaling of AI functionalities across the diverse B2B sales ecosystem.
Enhancing Discovery with Agentic AI in B2B
Beyond transactional efficiencies, agentic AI is also transforming product discovery for both retailers and B2B organizations. This extends to the capabilities powering advanced large language models (LLMs) like OpenAI’s ChatGPT, Google Gemini, and Perplexity. In an agentic AI-driven environment, discoverability is paramount for success. Do Forno highlighted that B2B companies can leverage AI agents to provide more than just basic customer service responses. With the right underlying data architecture, these agents can assist buyers in constructing complex orders by understanding nuanced product relationships and use cases.
For instance, a B2B buyer might need to assemble a component that requires a dozen individual products. If a supplier has not meticulously documented these relationships or provided clear content that links these products, buyers may struggle to find them. Agentic AI can bridge this gap by utilizing rich, contextualized product information. Suppliers are encouraged to provide detailed scenarios and use cases for their products, which can then be fed into AI models. This allows agents to better understand how products are used together, enabling them to respond to more complex, long-tail queries. Associating products with known experts or specifying their suitability for particular outdoor applications, for example, can significantly enhance an AI agent’s ability to recommend relevant items.
"The visibility is actually now way more complex – the GEO of it all," do Forno remarked, referring to generative engine optimization, a concept that builds upon traditional search engine optimization (SEO). This complexity underscores the initial steps many companies must take to even be considered for inclusion in AI-driven search results.
SEO Versus GEO: Optimizing for AI Discovery
The evolution from Search Engine Optimization (SEO) to Generative Engine Optimization (GEO) represents a critical shift in how businesses approach online discoverability, particularly in the age of AI. While SEO traditionally focuses on keywords to rank content in search engine results, GEO extends this by optimizing content for LLMs, aiming to surface results that demonstrate deeper contextual understanding.
"What’s different is you need to understand the intent versus just the keywords," do Forno emphasized. He elaborated that a sole focus on keywords can be superficial, akin to "a game." Instead, businesses require individuals who can interpret and articulate user intent in a multifaceted manner, going beyond simple product specifications. This involves connecting the dots between various pieces of information, not just the product data itself.
Consider a scenario where a buyer is undertaking a large project, such as building a house. This project entails numerous interconnected products. If a marketing team has only implemented basic SEO practices and has not comprehensively cataloged the company’s offerings, there is substantial work to be done. Expanding product information to include diverse build types and use cases is essential for AI agents to effectively guide buyers.
Furthermore, B2B product data presents unique complexities compared to B2C. Regulatory standards play a significant role. For example, if a material has specific legal restrictions on its use or handling, this information must be meticulously incorporated into product data, making it "super complex," according to do Forno.
The nature of buying channels also introduces another layer of differentiation. B2B buyers may be restricted from accessing certain products due to internal approval processes or procurement policies. This necessitates that product data accurately reflects these access constraints.
Finally, "fitment" is a crucial factor. A single product can have dozens of permutations based on attributes like color, size, and shape. How effectively a company categorizes and describes these variations directly impacts the frequency with which an AI agent will be able to surface that product in response to a buyer’s query. Accurate and detailed fitment data is therefore indispensable for maximizing product visibility in an agentic AI environment.
The Broader Implications for E-commerce
The divergent paths of B2B and B2C in agentic AI adoption signal a future where customer experiences will be increasingly personalized and automated. For B2C, the anticipation is that AI agents will handle a wider array of customer interactions, from product recommendations and personalized shopping experiences to post-purchase support, all with a higher degree of autonomy. This is driven by the generally simpler transaction structures and more standardized customer journeys in B2C.
In contrast, the B2B landscape, with its intricate supply chains, complex negotiation processes, and diverse regulatory environments, presents a more challenging, albeit potentially more rewarding, frontier for agentic AI. The current focus on solving specific pain points, such as reordering and PO processing, represents the initial phase of a long-term integration. As B2B companies build more robust digital foundations and gather more sophisticated data, the capabilities of their AI agents are expected to expand significantly. This could include AI-powered contract analysis, predictive inventory management, and even automated negotiation support for complex deals.
The Deloitte insights suggest that a strategic, iterative approach is crucial for B2B success. Companies that invest in core commerce infrastructure and focus on enriching their product data with contextual, intent-driven information will be best positioned to capitalize on the potential of agentic AI. The shift from keyword-centric SEO to intent-focused GEO is not merely a technical adjustment but a fundamental reorientation of how businesses present themselves and their offerings in an AI-driven marketplace. As agentic AI continues to mature, its impact on efficiency, customer satisfaction, and competitive advantage across both B2B and B2C e-commerce sectors will undoubtedly be profound.
