February 16, 2026

B2B agentic commerce stopped being theoretical this week. Oracle embedded AI agents across its Fusion Cloud Supply Chain and Manufacturing suite, unveiled February 10 at Oracle AI World in Mumbai. These are not concept demos. They are named agents with defined workflows, running inside a production ERP stack. One identifies requisitions eligible for competitive bidding, prepares sourcing events, invites suppliers, and sends notifications according to company policy. Another consults service history to find the right replacement part for a customer issue, then automatically places the order. A third reads PDF purchase orders, converts them into structured sales orders, and learns from user feedback to improve accuracy over time. Oracle bundled all of this at no additional cost to existing Fusion customers. That pricing decision matters as much as the technology. When agentic capabilities ship free and default, adoption stops being a strategic choice and becomes an inevitability.

The same week, Didero raised a $30 million Series A to bring AI agents to global supply chains. Where Oracle built agents deep inside its own stack, Didero takes the opposite approach. Its agents wrap around whatever messy infrastructure already exists. Procurement teams today "manage thousands of supplier interactions across email, ERP systems, and spreadsheets, much of it still handled manually." Didero embeds agents directly into those channels, reading and responding to supplier emails, updating ERP statuses, creating orders, and handling exceptions. "Within weeks of integration, these agents take on core procurement workflows like supplier communication, order tracking, and exception handling." Oracle offers a deeply integrated single-vendor agent stack. Didero offers system-agnostic agents that layer on top. Both approaches are finding customers.

Cathay Capital published an analysis today titled "Agentic AI Is a Massive Opportunity for B2B Software", drawing on meetings with AI startups, hyperscalers, and SaaS incumbents. Goldman Sachs' partnership with Anthropic stood out. The bank deployed AI agents for accounting and client onboarding, with their CIO describing the agents as "digital co-workers for process-intensive professions." That is not a commerce deployment specifically. But procurement, order management, and supply chain back-office work are process-intensive professions too. When a tightly regulated institution like Goldman embeds agentic AI into core operations, it validates the entire category beyond chatbots and copilots. Cathay's thesis is that established SaaS vendors sit at the center of this transition, shipping embedded, verticalized agents as core modules rather than add-ons. Startups like Didero innovate on cross-system agents for the gaps between platforms.

McKinsey published what reads as its most direct statement on the category yet, titled "The Agentic Commerce Opportunity". The framing captures a shift that previous coverage has circled without naming. Commerce is moving from responsive to predictive. AI agents do not wait for search queries. They infer needs and act, adjusting reorder frequency when a family's detergent usage increases during school months, or suggesting holiday purchases before the customer thinks to look. McKinsey projects agentic commerce will reshape $1 trillion in US retail revenue alone by 2030. Consumer-side projections are familiar territory by now. What matters more this week is the convergence on the B2B side, where Oracle, Didero, and Goldman Sachs are simultaneously placing large bets.

The startup ecosystem is thickening fast. George Krasadakis' February 2026 guide to agentic commerce catalogues over 90 startups building in the space, backed by firms like Andreessen Horowitz, Sequoia, and Accel. The landscape spans consumer agents from OpenAI, Google, Amazon, and Perplexity through infrastructure plays from Stripe, Visa, and PayPal. Krasadakis introduces a concept that deserves more attention than it is getting. Generative Engine Optimization, or GEO, is positioned as the successor to SEO. "Just as SEO optimised for Google, GEO optimises product catalogues for AI agents." With AI now accounting for up to 25% of referral traffic for some retailers according to Bain and Similarweb data, the optimization target has shifted from search engine crawlers to autonomous purchasing agents. Retailers who built their competitive advantage on Google rankings are discovering that a different kind of machine now decides which products get recommended.

SAP laid out its supply chain trends for 2026, centering agentic AI and multi-agent orchestration as the defining technical shift. The same Oracle announcement introduced AI Agent Studio, a platform for building, testing, and deploying agent teams across the enterprise, less a product feature than a meta-agentic development environment. LTIMindtree published a Cognitive Commerce Agents whitepaper describing how agent systems will restructure enterprise commerce from order intake through fulfillment. The enterprise vendors are not adding agents to their products. They are rebuilding their platforms to assume agents are the primary users.

B2B agentic commerce was, until recently, the quieter sibling of consumer-facing agent shopping. Not anymore. Oracle shipped named agents inside a production ERP. Didero raised $30 million to agent-wrap existing procurement channels. Goldman Sachs deployed Anthropic agents into core financial operations. Consumer agentic commerce still makes better headlines, but the back office is where purchasing decisions worth trillions actually happen. The agents are already there.