Summary of "Agentic AI in Retail" with Dan Rosado, Furniture.com

The Agile Brand with Greg Kihlström® — Episode #722
August 22, 2025

Episode Overview

This episode explores the evolving role of agentic AI in retail, featuring an in-depth discussion with Dan Rosado, General Manager at Furniture.com. Hosted by Greg Kihlström at Etail Boston, the episode delves into how Furniture.com is leveraging agentic AI to create a scalable, data-rich, customer-centric marketplace. The focus is on real-world implementations, challenges, best practices, and the organizational shifts required to embrace agentic and large language model (LLM)-optimized AI architectures in omnichannel retail.

Key Discussion Points & Insights

1. Furniture.com’s Unique Business Model

[02:12]

• Aggregating Brands with Trust: Furniture.com is positioned as a trusted aggregator, bringing together reputable furniture brands into a “one-stop-shop,” establishing consumer confidence.
• Scale and Complexity: The site features 65 retailers, 2 million SKUs, and 40,000 distinct pages, leading to "close to 80 billion combinations of ways that we can show product on our site." (Dan Rosado, [03:20])
• Marketplace Expertise: Dan Rosado’s background (homes.com, apartments.com, Cvent) informs a focus on scalable, high-confidence shopping experiences.

2. The Motivation for Embracing AI

[03:04]

• Limits of Manual Merchandising: With vast SKU and vendor diversity, manual/point-to-point integration is unmanageable and unscalable.
• AI as a Necessity: "The old ways... just don’t work. You need some sort of scalable way to make it look and feel curated." (Dan Rosado, [03:20])

3. What Is Agentic AI?

[04:27]

• Contrast with Traditional Automation: Traditional integrations require brittle, static mapping between systems.
• Agentic AI Defined: Agents are given intent and context, not rigid instructions. For example: “Instead of saying exactly where [a review] is... you basically tell the agent, go find me reviews on this product.” (Dan Rosado, [04:27])
• Adaptability: Agents can dynamically respond to changes in partner sites or feeds.

4. Real-World Agentic AI Use Cases at Furniture.com

[05:54]

• Product Enrichment Agent: Validates and augments basic product feeds by crawling partner sites (with permission) to find reviews, care instructions, videos, and other rich media.
  • "We’re able to then bring over to Furniture.com and make the shopper experience way better." (Dan Rosado, [05:54])
• Automated Distributed Transaction Agent: Enables cross-retailer checkouts (e.g., sofa from one retailer, rug from another in the same cart), with AI agents executing purchases on users' behalf.
  • “Behind the scenes we’ll actually use agents to go make those purchases on the shopper and Furniture.com’s behalf.” (Dan Rosado, [07:32])

5. LLM-Optimized Architecture and Its Importance

[10:26]

• SEO Parallels: As Google transformed the web for search, retailers now must optimize content for LLMs (e.g., ChatGPT, Perplexity).
• MCP Server Protocol: Use of new OpenAI protocols (e.g., MCP Server) to structure data for efficient LLM ingestion.
  • “You want all the LLMs to be able to read your site very quickly so they can get information quick.” (Dan Rosado, [10:26])

6. Advice for Smaller Retailers

[12:03]

• LLM Optimization Scales Down: Even if you only have 2,000 products, being LLM-friendly helps you appear in AI-driven shopping experiences.
• Agentic Data Augmentation: Retailers should use agents to supplement product info from vendors, not just rely on often incomplete or archaic feeds.

7. Organizational & Mindset Shifts for Agentic AI

[13:50]

• Addressing “Paralysis” with Action:
  • Initial hesitation is common due to the pace of AI evolution.
  • Solution: “Let’s give everybody in the whole company training.” (Dan Rosado, [13:50])
• Universal Training and Hackathons:
  • “We did a half day session [with OpenAI], then hackathons… That was step one.”
  • Progression from ChatGPT to code-level AI tools like Cursor AI, with all staff (devs/non-devs) trained and engaged in building prototypes.
  • "Just giving access to the tools, and training people up, and then doing actual events… has been really successful for us." (Dan Rosado, [14:48])

8. The Future of Agentic AI in Retail (5–10 Years)

[15:51]

• “AI-Native” Shopping Experiences:
  • Agents could automate purchases, but consumers will still want personalization and discovery.
  • "I still think that there’s that sense of discovery that shoppers… want to participate in." (Dan Rosado, [15:51])
• Customizable Commerce UIs:
  • Consumers may soon describe (in natural language) their ideal shopping experience (“make it look like Apple.com with a cart like Best Buy and fill it with products from Furniture.com”), with LLMs dynamically creating the interface.
  • “The shopper will tell and describe how they want to shop.” (Dan Rosado, [16:44])
• Retailer Imperative:
  • Must ensure their data is LLM-friendly so that they're visible and accessible in these customizable consumer interfaces.

9. ETail Boston Conference Insights

[18:04]

• Validation Through Community:
  • Face-to-face conversations reveal shared industry challenges.
  • “It feels good that we’re sort of all kind of going through this technology revolution together.” (Dan Rosado, [18:44])

10. Staying Agile as a Leader

[18:57]

• Constant Learning & Engagement:
  • Reading, networking, and direct engagement with younger employees.
  • “I try to have one-on-ones with... as close to 75 employees… and stay engaged with younger people and it keeps me sharp.” (Dan Rosado, [19:21])

Notable Quotes & Memorable Moments

• On Agentic AI’s Core Value:
  “What we mean by agentic AI... is instead of mapping it precisely, you can tell the agent what your intent is and what the context is... The agent can figure it out and understand the intent of what it is you’re trying to do.”
  (Dan Rosado, 04:27)

• On Product Enrichment:
  “We’re using an agent to not only go back to the websites of our partners… to validate basic information, but also pull additional information... care instructions, videos, any additional rich data.”
  (Dan Rosado, 05:54)

• On LLM Optimization:
  “You want all the LLMs to be able to read your site very quickly so they can get information quick... there’s an additional layer now... recommended to websites to create [specific structures] in order to optimize the experience.”
  (Dan Rosado, 10:26)

• On Getting Started with AI:
  “Just giving access to the tools and training people up and then doing actual events where they’re somewhat forced to use them... has been really successful for us.”
  (Dan Rosado, 14:48)

• On the Future of Retail UX:
  “Shoppers will create their own user experience through the LLM... The shopper will tell and describe how they want to shop.”
  (Dan Rosado, 16:44)

• On Leading with Agility:
  “I just try to stay engaged with younger people and it keeps me sharp.”
  (Dan Rosado, 19:21)

Key Timestamps

• 01:08 — Introduction to Agentic AI in retail
• 02:12 — Furniture.com’s model & Dan’s background
• 03:20 — Scope of SKU/catalog complexity
• 04:27 — Explanation of agentic AI vs. traditional approaches
• 05:54 — Real-world AI agent examples: data enrichment, distributed checkout
• 10:26 — Optimizing content for LLMs (SEO ≠ LLM optimization)
• 12:03 — Applicability for smaller retailers
• 13:50 — Mindset and process for organizational AI adoption
• 15:51 — Predictions: future AI-native shopper experience
• 18:04 — Insights from Etail Boston
• 18:57 — Staying agile as a leader

Conclusion

This episode offers actionable insights into the practical implementation of agentic AI within retail, emphasizing that technology, data infrastructure, and organizational mindset must all evolve together. Furniture.com’s journey—training every employee on emerging tools, investing in rich data feeds for LLMs, and reimagining the future interface of retail—serves as a blueprint for brands looking to thrive in the next era of omnichannel AI-powered commerce.