Vince ✕ Salesforce
Vince is a contemporary luxury fashion brand known for refined, timeless clothing with a modern edge. With a focus on elevated essentials for women and men, Vince is built around sophistication, restraint, and craftsmanship. As Salesforce developed new applications for its Agentforce AI platform, Vince was invited to explore how this technology could support an elevated e-commerce experience, one that mirrored the personal attention of in-store shopping.
Background
Salesforce’s AI demos had traditionally focused on functional tasks like order tracking or FAQs. This project aimed to stretch that definition: could Agentforce power a digital shopping experience that felt as refined and intelligent as a luxury fashion associate?
MRM, in collaboration with partner agency Rafter1 (later Acxiom), led a proof of concept that reimagined what AI could look like for luxury retail. I led the experience design spanning UX, brand integration, and content direction, shaping an AI assistant that reflected Vince’s tone while solving genuine customer needs.
Free form discovery & exploration
Tangibility of materials
Facets of luxury
in-store shopping
Personal service & advice
Ease of Transactional needs
Insight
A simple chat interface wouldn’t capture the tactile, emotional nature of luxury fashion. We needed a conversational experience that was visually rich, brand-true, and adaptable across different styling journeys.
Our solution combined:
Intelligent product search (direct queries).
Mood-driven prompts (“What’s right for a relaxed dinner in Malibu?”).
Multi-item outfit recommendations (styling journeys).
To support this, we enriched Vince’s product data with AI-derived attributes such as silhouette, fabric texture, and sophistication. Models evaluated copy, metadata, and product imagery to produce ranked, nuanced insights. This meant the assistant could respond in a more human voice:
“These shoes pair well with your outfit, they share the same refined texture and muted palette.”
AI-generated textures and overlays were surfaced directly inside the chat interface, giving users a richer sense of fabric detail without disrupting product detail pages.
Process and Governance
To ensure brand fit, we developed prompt templates and scoring frameworks that tuned the large language models to Vince’s luxury tone. Internal validation loops scored responses for consistency and relevance, replacing traditional metrics in this proof of concept stage.
Inclusivity was handled by enabling user-driven product variations allowing exploration of different fits, tones, and looks without prescriptive styling.
Headless Commerce (AI) Architecture
Conversational UI
AI orchestrated journeys, with intelligent context and not just chats
APIs
Modular and performant at Scale
Real time catalog, inventory, pricing, promotions
Agents
Agent Instructions, Descriptions, and Scope defined in Topics
Heavy reuse of Salesforce Platform components (e.g. Flow, Apex)
Trust layer
Ensures safe, compliant and enterprise ready use of GenAI
LLMS
Adaptive learning models to reflect seasonal and event-driven personalisation
Data cloud + metadata
Retrieval Augmented Generation (RAG) Support
Unstructured (web) and Structured Data Support
Result
The demo was a major success. It impressed both Salesforce and Vince stakeholders, reframing Agentforce not just as a service assistant but as a persuasive digital stylist.
For Salesforce, it provided a compelling new sales narrative for Agentforce.
For Vince, it showed the potential of generative UX to extend luxury brand storytelling into digital channels.
Good stuff
This project fused AI data enrichment, conversational design, and generative imagery into a single luxury proof of concept showing that AI can be emotionally intelligent, not just transactional.
AI E-commerce POC
Date: 2025
Company: Vince (in collaboration with Salesforce)
Project: Agentforce Agentic AI E-commerce Proof of concept
Role: Experience Design, Concept Strategy, UX Design, Gen AI Content production
Team:
Gareth Osborne (CTO)
Graeme Mulvaney (Technical Director)
Matt Gardner-lewis (Account Manager)
Elliot Herbert (Application Architect)
Tracy LaPlante (Account Director)