AN FRANCISCO — January 10, 2026 — Salesforce is expanding its Agentforce 360 platform to address what it describes as a growing problem in AI-led retail: companies deploying multiple autonomous agents that work well inside individual systems, but fail when customers move across marketing, commerce, and service interactions.
In a company blog post, Salesforce Commerce and Retail Clouds executive Nitin Mangtani argued that many retailers are building agents “one function at a time,” resulting in disconnected experiences when campaign promises, product recommendations, fulfillment expectations, and support responses are generated from different systems with limited visibility into one another. “Each agent may be doing its job, but it’s doing it in isolation,” the post said.
“Context” moves from integration project to platform claim
Salesforce’s central message is that “agentic commerce” breaks down without a shared context layer that can be consistently interpreted and enforced—particularly around inventory, order status, customer history, permissions, and governance.
The company positioned do-it-yourself approaches as costly and difficult to govern at scale, describing a pattern where retailers attempt to connect campaign data, browsing behavior, catalog and inventory systems, order history, and service cases through custom APIs and pipelines—then struggle with drift, latency, and inconsistent permissions enforcement over time.
Agentforce 360, Salesforce said, is intended to serve as an “enterprise-grade context engine” spanning sales, commerce, marketing, and service. The company described it as a “shared intelligence layer” where multiple agents can “reason” using the same context rather than relying on system-by-system logic.
New capabilities highlighted for commerce and messaging
Salesforce pointed to several commerce-related capabilities it says are now operating on that shared foundation:
- Two-way messaging across channels (including SMS and WhatsApp today, with email planned), positioned as a way to convert one-way notifications into interactive shopping and service conversations. Salesforce
- Contextual Search, described as intent-aware search for both site search bars and shopping agent logic; Salesforce said it is in open beta. Salesforce
- Agentforce Commerce Guided Shopping updates, which Salesforce said can proactively assist shoppers on a retailer’s site by confirming inventory, calculating shipping, and answering product questions within a single conversational flow. Salesforce
Salesforce said Agentforce Commerce Guided Shopping, Contextual Search (open beta), Two-Way SMS, and Two-Way WhatsApp are available now, while Two-Way Email and “Marketing and Commerce Connected Journeys” are expected in February.
Salesforce ties agent adoption to holiday performance metrics
To support its argument that agents are becoming embedded in retail operations, Salesforce pointed to its own holiday shopping analysis. In a separate Salesforce report published January 8, the company said retailers deploying branded AI agents saw a “59% higher growth rate” on average (6.2% year-over-year sales growth versus 3.9% for others in its comparison set) and that shoppers used retailers’ AI and agents for customer service 126% more during the holiday rush than in the prior two months.
Salesforce’s holiday analysis was based on aggregated commerce activity across more than 1.5 billion shoppers in 89+ countries, and it noted that its benchmarks are not indicators of Salesforce’s own financial performance.
Broader push: agents, checkout protocols, and third-party AI surfaces
Salesforce’s “shared context” framing arrives as more commerce activity begins to originate from AI-powered discovery surfaces outside retailers’ owned sites, increasing pressure on brands to manage identity, permissions, and transaction integrity across channels they do not fully control.
Salesforce has been building toward that environment. In October 2025, the company announced support for an “Agentic Commerce Protocol” initiative with Stripe and OpenAI aimed at enabling “Instant Checkout” flows inside conversational interfaces while merchants retain control of fulfillment and customer relationships, according to Salesforce.
The company has also emphasized external model partnerships as part of its agent platform strategy. Reuters reported in October 2025 that Salesforce expanded ties with OpenAI and Anthropic to enhance Agentforce capabilities across its products, including commerce-related features that could facilitate purchases via ChatGPT.
Industry-wide, questions about security, account access, and data rights are becoming more visible as agents become more action-oriented. Reuters reported in November 2025 that Amazon sued Perplexity over an “agentic” shopping tool, reflecting rising tensions over how automated agents interact with consumer accounts and merchant platforms.
For now, Salesforce’s latest post reads as both a product message and a platform thesis: as retailers deploy more agents, the company wants the differentiator to be less about “having an agent,” and more about whether agents share governed context—so customers don’t experience the seams between marketing, shopping, and support. (Because nothing says “brand trust” like your storefront confidently recommending the thing the customer just returned.)
