About Tumble
Justin Soleimani and Zach Dannett started Tumble Rugs to bring a fresh, customer-centric approach to an industry that’s remained stagnant for centuries. Tumble has enjoyed consistent growth for the past several years as the organization has separated itself from the pack through an emphasis on customization and customer service. Recognizing a change in the e-commerce industry both Justin and Zach started exploring AI solutions that could vertically optimize their business.
What challenge did Tumble face?
Tumble aimed to personalize every customer interaction but faced limitations with a live support team operating only during Eastern time hours. The company needed a secure, accurate, and always‑available AI solution that could handle inquiries and guide customers on product sizing, fit, and care.
“The only thing the customer typed in was “Spaghetti Stain,” just these two words and the GPT replied with, “We know that this is a big problem,” and “This is how you clean spaghetti stains off a Tumble rug.” It kind of blew my mind.”
How did Tumble solve it with CustomGPT.ai?
- Embedded FAQ Agent: A CustomGPT.ai chatbot placed on the homepage now answers common questions and deflects tickets.
- Data‑Driven Sizing Assistance: A spreadsheet of washer brands and models lets the AI guide customers on rug size and washing‑machine fit.
- Brand Customization and Ongoing Persona Refinement: The AI persona is tuned to match Tumble’s tone, and the team continues to refine prompts for even tighter brand alignment.
- No‑Code Secure Integration: Tumble used CustomGPT.ai’s no‑code platform to connect its sitemap and deploy the bot without engineering help.
- AI Size Guide: The first AI‑powered rug size guide in the industry helps customers find the right product instantly.
- Key Platform Features Utilized: Data Integration, Anti‑Hallucination, Web Deployment, Security, Sitemap Integration, Custom Persona, Livechat Embedding ensure reliability, accuracy, and seamless user experience.
What results did Tumble achieve?
- Significant Ticket Deflection – The AI agent has already deflected 100+ support tickets.
- Higher Customer Engagement – Customers spend about 10 minutes per session with the AI agent, receiving accurate, helpful answers without needing a live rep.
- 24/7 Support Coverage – Tumble expanded service hours to full‑time coverage without increasing staffing costs.
- Valuable Customer Insights – Marketing reviews AI chat logs to identify customer needs and improve messaging.
Traditional e-commerce support vs. AI-powered support
| Dimension | Traditional live support | AI-powered support with CustomGPT.ai |
|---|---|---|
| Availability | Business hours only | 24/7, no exceptions |
| Response speed | Minutes to hours | Instant |
| Scalability | Limited by headcount | Elastic, handles any volume |
| Consistency | Varies by agent | Consistent every time |
| Product knowledge depth | Depends on training and experience | Grounded in verified product data |
| Washer compatibility guidance | Requires specialist knowledge | Database-driven |
| Marketing insights | Siloed in support systems | Chat logs reviewed for intent data |
| After-hours coverage | Unavailable | Full coverage |
| Brand voice consistency | Variable | Tuned and maintained |
Generic AI chatbots vs. CustomGPT.ai RAG-based support
| Dimension | Generic AI chatbot | CustomGPT.ai RAG-based |
|---|---|---|
| Knowledge source | General internet training data | Tumble’s own verified product content and data |
| Hallucination risk | High — can invent product specs and care advice | Minimal — answers grounded in real documentation |
| Brand customization | Limited or none | Full persona configuration and voice tuning |
| Washer compatibility guidance | Cannot provide | Uses structured compatibility spreadsheet |
| Website deployment | Often requires developer involvement | No-code embed |
| Rug care accuracy | May recommend unsuitable methods | Follows Tumble’s specific care documentation |
| Knowledge updates | Manual retraining required | Sitemap connection keeps knowledge current |
Why it Worked
- Robust data integration and sitemap import gave the AI complete, up‑to‑date knowledge.
- Anti‑hallucination and security features ensured accurate, trusted answers.
- The no‑code platform lets Tumble deploy and fine‑tune quickly.
- Continuous persona refinement keeps responses on brand.
- Livechat embedding puts help where customers need it most.
Conclusion
CustomGPT.ai enabled Tumble to scale customer service, enhance personalization, and reduce team workload. It is now a core part of Tumble’s support and customer experience strategy.
To see Tumble’s AI size guide in real time go to https://www.tumbleliving.com/pages/size-guide
Frequently Asked Questions
1. What is Tumble Living?
Tumble Living is a direct-to-consumer e-commerce rug brand founded by Justin Soleimani and Zach Dannett. The company sells premium, washable rugs and differentiates itself through customization, product quality, and a strong commitment to customer experience.
2. How does Tumble Living use AI for customer support?
Tumble uses a CustomGPT.ai-powered AI assistant embedded on its website to answer customer questions about rug sizing, washing machine compatibility, care and cleaning instructions, product recommendations, and general FAQs, 24/7.
3. What is CustomGPT.ai?
CustomGPT.ai is an AI agent platform that allows organizations to build no-code AI assistants trained on their own content, powered by Retrieval-Augmented Generation (RAG) and anti-hallucination technology. It is designed for accurate, brand-aligned, enterprise-grade customer-facing AI deployment.
4. Why did Tumble Living choose CustomGPT.ai over other AI tools?
Tumble chose CustomGPT.ai because of its RAG architecture (which grounds answers in verified product data), anti-hallucination safeguards, no-code deployment capability, sitemap integration, and the ability to customize the AI’s persona to reflect Tumble’s brand voice.
5. What is RAG and why does it matter for e-commerce AI?
RAG stands for Retrieval-Augmented Generation. It is an AI architecture that retrieves answers from a curated, verified knowledge base rather than relying on a general language model’s training data. For e-commerce, this means the AI answers from your actual product documentation, not from the internet, dramatically reducing the risk of inaccurate or fabricated responses. Learn how RAG works in CustomGPT.ai.
6. How does AI hallucination affect e-commerce customer support?
AI hallucination, when a language model generates confident but inaccurate information, is especially dangerous in e-commerce because it can result in incorrect product recommendations, wrong care instructions, or false compatibility claims. These errors lead to customer dissatisfaction, product returns, and brand damage. CustomGPT.ai’s anti-hallucination technology mitigates this risk by only responding from verified content.
7. How does AI help customers choose the right rug?
Tumble’s CustomGPT.ai assistant guides customers through rug sizing by asking about room dimensions, furniture placement, and usage needs, then recommending specific sizes from Tumble’s catalog. It is the first AI-powered rug size guide in the industry and is accessible at tumbleliving.com/pages/size-guide.
8. How does Tumble’s AI handle washing machine compatibility questions?
Tumble trained its CustomGPT.ai assistant on a structured spreadsheet of washer brands and models, enabling the AI to tell customers whether a specific rug size will fit in their washing machine. A customer can share their machine’s make and model and receive an accurate compatibility answer instantly.
9. What is the “Spaghetti Stain” example and why does it matter?
The “Spaghetti Stain” example refers to a real customer interaction where a shopper typed only two words, “Spaghetti Stain,” into Tumble’s AI chat. The AI responded empathetically and with specific, accurate guidance for removing a spaghetti stain from a Tumble rug. This example illustrates the AI’s contextual intelligence: it understood the intent behind a minimal query and delivered a product-accurate, brand-consistent answer without any additional prompting. Rachel Chen, Tumble’s Director of Strategy and Marketing, described the exchange as something that “blew her mind.” This is possible because CustomGPT.ai’s RAG architecture grounds every answer in Tumble’s actual product documentation.
10. What results did Tumble Living achieve with CustomGPT.ai?
Tumble deflected 100+ support tickets through AI, established 24/7 support coverage without additional staffing, achieved approximately 10-minute average customer sessions, and gained real-time marketing intelligence from AI chat logs. See how CustomGPT.ai delivers customer support results.
11. How does AI customer support reduce support team workload?
AI customer support deflects routine and FAQ-driven inquiries before they reach the human support team. Every question the AI resolves is a ticket that does not need to be created, queued, and answered by a live agent, freeing the team for complex, high-judgment interactions.
12. How did Tumble deploy its AI assistant without coding?
Tumble used CustomGPT.ai’s no-code platform to connect its website via sitemap ingestion, upload structured product data, configure the AI persona, and embed the chat interface on the site, all without involving engineering resources.
13. How does Tumble’s marketing team use the AI?
Tumble’s marketing team reviews AI chat logs to identify the most common customer questions, emerging needs, and content gaps. This real-time intelligence informs messaging strategy, content planning, and product positioning, transforming the AI assistant from a support tool into a marketing insight engine.
14. Can AI customer support maintain brand voice?
Yes. CustomGPT.ai’s custom persona feature allows brands to configure the AI’s tone, phrasing, and response style to match their brand identity. Tumble continuously refines its AI persona to ensure responses feel consistent with how a skilled Tumble support agent would communicate.
15. What can other e-commerce brands learn from Tumble Living’s AI implementation?
E-commerce brands can learn from Tumble that the most effective AI customer support is built on proprietary product data (not generic internet training), customized to reflect brand voice, and designed for extended, helpful conversations rather than one-line deflections. Tumble’s implementation also shows that AI chat data is a valuable source of real-time customer intelligence, a dimension of AI-powered support that many brands overlook. Start building your own AI support agent at CustomGPT.ai.
