Enterprise AI Agents Move from Experiments to Everyday Work
Enterprise AI agents are AI-driven software components that can understand instructions, interact with business systems, and take actions to automate routine ecommerce and procurement tasks across B2B and B2C operations. Instead of only answering questions, these AI commerce agents are now embedded directly into order flows, catalog management, customer conversations, and spend management tools, so they can complete work like placing orders, updating records, or flagging policy issues. The latest releases from Salesforce and Amazon Business show how quickly this shift is happening in real enterprise environments. Salesforce is wiring AI agents into its Agentforce Commerce platform, while Amazon Business is expanding AI assistants and analytics for B2B procurement AI. Together, these moves mark a clear step from standalone chatbots toward agentic enterprise workflows where automation acts inside core systems, not around them.
Salesforce’s Agentforce Commerce Brings Agents to B2B and B2C
Salesforce has added AI agents to Agentforce Commerce for both B2B and B2C brands, extending ecommerce automation into everyday buying and merchandising tasks. On the B2B side, a Buyer Agent meets customers in WhatsApp or SMS and can orchestrate the procurement process without forcing a portal login or phone call. A buyer can describe a repeat order in natural language, and the agent confirms the SKU, contract pricing, and order details. The Merchant Agent helps internal teams manage catalogs in plain language, from sorting products to setting boost and bury rules. For B2C, Salesforce has introduced a Shopper Agent and agentic commerce search, alongside deep integrations with ChatGPT, Gemini, and Google AI Mode that connect product catalogs directly to conversational interfaces while keeping the retailer as merchant of record and ensuring every order lands in the same core commerce platform.
Amazon Business Targets Procurement Automation and Spend Governance
At its Amazon Business Exchange event, Amazon Business announced new AI tools aimed at B2B procurement AI, focusing on both workflow automation and stronger spend governance. The headline is Amazon Quick, an AI assistant that integrates with existing tools such as Slack and Microsoft Outlook to automate repetitive tasks, pull cost information, review supplier proposals, and generate reports from company data. Unlike simple chatbots, Quick is designed to take actions in context while still requiring user approval, reinforcing governance. Alongside this, Amazon Business is enhancing spend management tools with an upgraded Spend Visibility dashboard that offers a redesigned interface, near real-time data, access for more users, and 24 months of historical data. According to Amazon Business’s State of Procurement research, “73% of senior leaders believe stronger data and analysis capabilities will be critical to improving operations over the next two years.”
Core Adoption Challenges: Integration, Governance, Operational Readiness
Despite the appeal of ecommerce automation, enterprise AI agents face three consistent adoption challenges: integration, governance, and operational readiness. Integration means agents must connect to catalogs, pricing, inventory, orders, and procurement systems so actions are accurate and reconciled; Salesforce stresses that its agents are wired “natively into your catalog, inventory, and orders from the first day.” Governance requires clear spend controls, audit trails, and policy enforcement, which Amazon Business addresses with anomaly detection, expanded spend dashboards, and user approval flows in Quick. Operational readiness covers change management, training, and process design so teams trust the AI and know when to intervene. Procurement leaders also report pressure to do more with fewer resources, and nearly half say balancing efficiency with rising operational demands is their biggest challenge, making these three areas central to any enterprise AI deployment plan.
Toward Agentic Commerce and Procurement Workflows
Taken together, these updates signal a shift toward agentic enterprise workflows in commerce and procurement, where AI commerce agents do more than surface insights—they complete work. In ecommerce, agents now help customers place orders in natural language, support merchants with catalog tuning, and power discovery across channels. In procurement, assistants assemble spend reports, flag unusual buying patterns, and compare vendor terms against historical agreements, extending the reach of spend management tools. This fits a wider move away from transactional purchasing toward intelligent, data-driven decision-making. As platforms embed enterprise AI agents directly into their core products, the question for organisations is less about if they will adopt AI, and more about how they will design processes, policies, and teams so these agents can automate routine tasks while humans focus on strategic choices and exceptions.
