AI support for a high-touch retail environment
Curating the perfect product assortment from a catalog of more than 10,000 items is nearly impossible for a human. For generative AI and agentic AI, it’s effortless.
We’re used to seeing AI agents and GenAI chatbots deliver fast, personalized product recommendations in e-commerce settings. However, retailers that rely on physical experiences can face challenges implementing AI.
Take furniture sales. Most shoppers will adopt a hybrid approach, using online channels for research and inspiration, followed by high-value purchases in showrooms and stores (or vice versa). They want to feel the fabric, test the comfort, and talk it through with someone first.
Buying furniture for a room can be as personal and detailed as a styling session at a luxury department store. Say you like the couch, is there a chair that matches? What about an end table? What are the right dimensions for your room and what goes with your color scheme? What’s pet-proof?
That blend of tactile experience and high-touch service is why many furniture retailers, even digital-first ones, are still focused on expanding their brick-and-mortar presence. And it’s why iSoftStone’s client — a fast-growing furniture retailer backed by a private investment group — tasked us with using AI capabilities to support the workflow of their human sales associates, not replace it.
Generative AI is proving to be one of the most transformative technologies of our time, reshaping how businesses interact with customers, create value, and personalize experiences. In the highly competitive retail industry with showrooms, tapping into this opportunity has become essential.
VP of Business Intelligence & GenAI | Leading Furniture Retailer
Creating a cost-effective AI chatbot
Says Sicheng Peng, iSoftStone’s senior client delivery partner, “We are always excited to collaborate with clients on new innovations and bring wonderful experiences to their customers.”
The retailer’s team wanted a solid test case to prove the value and safety of AI in a customer-facing and sales setting. Any AI tool had to be rolled out incrementally, be lean and cost-effective, and, crucially, work for a seasoned salesperson as well as a new hire.
We created an AI chatbot that accesses our client’s extensive catalog and delivers targeted product recommendations based on customer preference and regional warehouse data.
In considering the right balance of AI model performance and cost for the client, we brought together:
- Azure AI Foundry as the engine
- Azure OpenAI/GTP-4o mini for natural language responses
- Azure AI Search for indexing our client’s product catalog
- Python/JSON for the catalog conversion
- User feedback loops to gather insights and fine-tune the experience.
To ensure excellent quality, we followed responsible AI practices and put the AI chatbot through rigorous QA and stress testing.
Finally, to make the AI chatbot a usable and meaningful AI tool in a physical retail environment, we integrated it with our client’s CRM.
A successful pilot phase launches AI in physical stores
Our client’s salespeople use tablets throughout their entire day, from clocking in and scheduling shifts to logging sales in the CRM and monitoring their performance ranking. It’s their first point of contact for everything that happens on the floor.
They don’t need to break that rhythm to get AI product recommendations. A user simply taps a floating icon in the corner of the CRM dashboard and chooses from pre-determined options, including:
- Customer type (individual, couple, family, etc.)
- Size of living space
- Priorities such as comfort, style, or durability
- Special considerations like pets or budget range.
A natural language AI prompt is created behind the scenes and the chatbot returns the product recommendations together with talking points, sales scripts, and a website link in case the salesperson needs to show the client images.
Salespeople can also adjust the delivery center or search based on zip code for regional warehouse availability and/or direct delivery to customers. Catalog inputs are refreshed and reindexed every night so users have the most accurate, up-to-date information at their fingertips.
Because the AI chatbot is capable of handling complex queries such as complete rooms of furniture, complementary items, and qualitative attributes, it takes on the cumbersome work out of scrolling, filtering, and manually cross-referencing product specs.
As the retailer’s VP of business intelligence and GenAI enthusiastically notes, “This innovative solution provides expert, persona-driven product advice in real time, helping associates deliver highly relevant messages that truly resonate with each customer.”
Sales teams are delivering richer in-store customer experiences — and more quickly — while simultaneously growing their knowledge of products and customer needs. And, when they’re not with customers, they can also use the tool to research winning sales strategies.
Thanks to iSoftStone’s end-to-end delivery — from strategy to technical implementation — our company not only accelerated time-to-value but also established a scalable foundation for expanding Generative AI into additional use cases across the business.
VP of Business Intelligence & GenAI | Leading Furniture Retailer
AI product recommendations drive conversions
More than 100 sales associates in 11 stores are already using the AI product recommendation chatbot.
The VP of business intelligence and GenAI confirms that results in those stores have been incredibly positive: “By reducing friction in the buying journey and making recommendations feel intuitive and personal, we are already seeing early lifts in conversion rates, average order value, and customer satisfaction.”
The retailer is now rolling out the chatbot, region by region, to each of its 200 stores across the country, with approximately 5-10 new users added daily. As part of the ongoing investment in the tool, iSoftStone’s team is collecting user feedback and data to refine the AI engine and ensure the tool continues to deliver an accurate and user-friendly experience as it scales.
Next up, an AI-powered support assistant
With one proven AI use case in hand, the client is excited to add more AI tools and solutions to its technology portfolio.
“This client trusts our skills,” says Khairul Nashran Nazari, an iSoftStone’s technical consultant who works closely with the retailer, “and it really gives us the motivation to keep on building stuff like this and exploring what AI technology can do for them.”
Together we’re building a second tool — an AI chatbot that empowers associates with fast, intuitive access to knowledge from teams across the company. It works with the same AI engine as the product recommender but with the Azure AI Search inputs coming from a broader range of formats including Word docs, PowerPoints, PDFs, and HTML files.
The goal is to help salespeople strengthen their tactics and deepen their policy and business operations understanding. If they want to understand how commissions are calculated or how to create delivery documents, they’ll be able to ask the chatbot and discover answers for themselves.
And, to keep user experience smooth, the support chatbot will be accessible from the same floating button as the product recommender — right there on the CRM dashboard.
AI made for humans, solutions for all sectors
An elegant, simple AI tool that does more, and does it leanly and safely. That speeds up what’s slow and clarifies what’s complex. That empowers people to do their jobs in the best way possible.
Our client needed AI experiences made for the real world. And we listened.
