The Dawn of AI-Native Shopping: Inside the Dreamforce 2025 Retail Keynote

The Dreamforce 2025 Retail Keynote, led by Alison Weinstein, Senior Vice President of Sales Digital at Salesforce, declared that the shift to AI-Native Shopping is already a reality. The session went beyond theory, showcasing how leading retailers are moving from simple AI tasks to becoming true Agentic Enterprises that transform both the top and bottom line.

The urgency was underscored by a startling statistic: over half of product discovery will happen on a Large Language Model (LLM) this holiday season, and over 21% of global sales will involve agents or AI in the process, a massive transformation in just 12 months.

The Agentic Retail Formula: Data, Conversation, Scale

Alison Weinstein and Michelle Collins laid out the three pillars required for retailers to thrive in this new AI-native world:

• Trusted Data Foundation: Agents must be entrusted with data in a safe way. The key is creating a deeply unified platform and a data foundation that powers identity experiences at a new level.
  

• Conversational and AI-First: Retailers must be available 24/7 and meet customers where they are, moving beyond static search and forms to personalized, two-way interactions.
  

• Productivity and Scale: Agents must integrate into retail workflows to give employees time back and scale operations, delivering consistent experiences across all touchpoints.
  

This new reality is powered by Agent Force 360, which is built on the philosophy that the customer journey must be owned by the retailer, especially the parts LLMs cannot replicate—conversion, retention, and loyalty.

The Data Activation Imperative

Michelle Collins introduced the pivotal role of data activation, emphasizing that retailers have an abundance of siloed data (POS, mobile, inventory) that must be unified to fully power agentic experiences.

• Data 360: This redefined platform is the key to activating data's full value. It harmonizes data with a "zero copy" approach, liberating trapped data from back-end systems to deliver a truly unified customer view.
  

• Action Across Channels: Once activated, this data can be actioned across any channel, powering marketing, commerce, and service. This foundation is essential for supercharging agents, enabling capabilities like two-way messaging with personal shoppers or stylists.
  

The End-to-End Agentic Journey: Northern Trail Outfitters (NTO)

A detailed demo featuring NTO showed the seamless journey of a customer, Lauren Bailey, moving from digital discovery to in-store pickup and service:

1. Conversational Commerce & Marketing Productivity

• Discovery: Lauren’s ad interaction on Instagram (fueled by personalized promotions using Data 360 insights) transitioned seamlessly to her laptop.

• Two-Way Chat: Lauren used the Agent Force Trail Guide on the website, asking natural language questions about tent setup, price range, and warranty. The agent not only answered but curated a selection of products and allowed her to add the item to her cart, showcasing intuitive, tailored, conversational shopping.
  

• Agent-Driven Marketing: The NTO team used a natural language prompt to ask Agent Force to create an email and SMS campaign for loyal customers. The agent drafted the entire campaign, including the flow of actions in Marketing Cloud Next. It then used Einstein Segment Creation to refine the target segment using insights from Data 360.
  

• Two-Way SMS: Lauren received the SMS campaign and initiated a two-way conversation with the agent, asking for waterproof jacket recommendations. She then instructed the agent to add them directly to her wish list, demonstrating an action-oriented marketing message.
  

2. The Store Transformation & Agentforce Voice

• Modern POS: In the store, the associate, Adam, used Salesforce Point of Sale, which unifies omni-channel orders, inventory, and clienteling. Adam immediately saw Lauren’s tent order, loyalty status, and her Agent Force-saved rain jacket wish list item. This single view allowed him to proactively inform her about her near-next loyalty tier and process a ship-to-home order for the out-of-stock item, turning a potential friction point into a loyalty moment.
  

• Agent Force Voice Saves the Sale: When Lauren’s plans changed, she used the Agent Force Voice option in the NTO app (no typing required). The voice agent proactively used Lauren’s collected trip data to suggest an immediate in-store pickup at a store near her new camping destination, saving a purchase that would have otherwise been a cancellation.
  

• The Connected Loop: Slack received an Agent Force notification summarizing the pickup context for the Klamath Falls store associates, ensuring the in-store team was ready for the sudden change, and updating Lauren’s full CRM profile in real-time.
  

Trailblazer Insights: Breitling and David Yurman

Breitling: Data Before AI

Rajesh Anmuga Sundaram, from Breitling, spoke about the brand’s four major transformations, including a heavy push toward Direct-to-Consumer (D2C). His main advice to the audience was crystal clear:

• Challenge 1: Data. "Please, please spend your time and money to consolidate the data before you even think about AI." He stressed that great data quality in one single place is fundamental.
  

• Human Coexistence: The luxury brand, which is based on emotional connection, believes AI will coexist with humans, not replace them.
  

• Exciting Use Cases: Breitling is co-developing a predictive website with Salesforce that dynamically builds its UI based on customer input, and has built a "Breitling GPT" for internal users to drive data-driven decisions instead of purely emotional ones.
  

David Yurman: The Human Touch in Conversational Commerce

Christian Fortucci, CTO of David Yurman, highlighted how the luxury brand is blending art and science by ensuring technology feels invisible.

• Agent Force Chat Success: By moving from a legacy menu-based system to a full conversational Agent Force chat, David Yurman achieved higher engagement, higher ticket deflection, and most proudly, higher customer satisfaction.
  

• Guiding Principle: They trained their bot to communicate in the brand voice to ensure every interaction felt authentic and trustworthy—no hybrid legacy approaches allowed.
  

• Generative Search Optimization (GSO): Fortucci emphasized the threat of customers transacting entirely within a third-party browser agent without ever visiting the brand's site. Retailers must develop a GSO plan to control the narrative, ensuring LLMs respond with the brand's approved images, copy, and links.

The keynote concluded by reinforcing that a unified, self-learning, customer-obsessed Agentic Enterprise is essential for thriving in the new retail landscape.