Crescendo Lab Blog
Blog |

Case Study: MR. LIVING Furniture Brand in Taiwan Turns AI Agent into a “Senior Sales Consultant” to Save 160 Hours Every Month

Kaya Heimowitz

Table of Contents

Share this article

Related Tags

In Taiwan’s home furnishing market, MR. LIVING is known for its furniture inspired by Nordic designs and is committed to maintaining a high-quality customer journey. By continuously advancing OMO (online–merge–offline) integration, the brand aims to deliver the best possible customer service experience.

However, as the brand actively expands its market and grows in a healthy way, more orders inevitably create bottlenecks—its LINE Official Account surpassed 16,000 messages per month, placing huge pressure on the customer service team.

“We receive a large volume of repetitive questions every day. Human agents are overwhelmed, yet complex needs still require manual handling.” 

 

— MR. LIVING | CTO, Lin Yu-Fan

This became the starting point for the brand to seek out Crescendo AI—not just to answer questions, but also to intelligently route conversations and support sales conversion.

Through working with Crescendo Lab, MR. LIVING implemented the enterprise-grade conversational AI platform CAAC and its AI Agent solution. During the testing period, AI was already able to handle routine text messages and was estimated to save the team 160 hours of repetitive labor each month—turning every customer interaction into an opportunity for brand growth.

The Challenge for Human Customer Service: When High Engagement Becomes High Pressure – The Hidden Cost of Service

門市情境照3

Repetitive Questions Flood Human Agents, Turning Customer Service into “Information Middlemen” Instead of “Service Providers”

Before introducing AI, MR. LIVING analyzed over ten thousand LINE conversation records and found that the team’s time was heavily compressed by repetitive questions—86.8% of the first messages sent by customers were in text form, and 43% of them were basic inquiries about store information, invoices, payments, warranties, and deliveries.

Although these questions were simple, they consumed nearly half of customer service working hours. With over 16,000 messages to handle every month, agents were constantly replying to FAQs, checking orders, and copying templates. Their time was drained by these “mechanical conversations.”

Over time, the role of customer service shifted from “relationship builders” to mere “information intermediaries,” leaving little room to truly understand customer needs or actively guide them toward a purchase.

This not only affected service quality but also weakened the team’s sense of achievement and professional value.

“We don’t want our customer service agents to be trapped in a loop of answering FAQs and have no time to understand what customers really need.”

 

— MR. LIVING | CTO, Lin Yu-Fan

This sentiment reflects not only MR. LIVING’s reality before adopting Crescendo’s solution, but also the common anxiety many traditional industries face at the early stages of digital transformation.

Communication Black Holes Across Departments: Fragmented Roles and Inconsistent Customer Experience

Beyond the burden of repetitive questions, a more challenging issue lies in cross-functional collaboration. In MR. LIVING’s chat logs, 57% of customer inquiries fall under more advanced needs, such as order modification, repair arrangements, promotional partnerships, or style-matching suggestions.

These questions often involve three departments: customer service, sales, and design consultants. Under traditional workflows, messages had to be passed along step by step, forcing customers to repeat their issues and wait for responses—naturally leading to a poorer experience.

A deeper challenge comes from misaligned department goals:

  • Customer service is evaluated on response speed
  • Sales focuses on conversion performance
  • Design consultants care about overall styling satisfaction

Without a shared platform and integrated data, these three workflows are hard to align, resulting in gaps in the customer journey.

This kind of “functional black hole” is especially common in high-interaction industries. What appears to be a single conversation actually spans multiple touchpoints—and this was precisely one of the key pain points MR. LIVING wanted to address.

Sensitive Issues Require Human Intervention: Collaborating with AI Means Balancing Efficiency and Trust

In addition to efficiency, the customer service team also guards brand trust. After-sales service in the furniture industry often involves high transaction amounts and strong emotions—such as returns, delays, warranty claims, and repairs. If these are mishandled, they can quickly escalate on social media.

That’s why, even before bringing in Crescendo AI, MR. LIVING had already set a clear principle. While they value the application of AI tools and expect AI to significantly boost service quality and efficiency, they take a cautious stance: human agents must remain the final gatekeepers.

They thus defined an operating model of:

“AI handles routing and information structuring; humans handle judgment and reassurance.”

This principle challenges the common misconception that AI is purely a replacement for humans. Instead, AI becomes the brand’s assistant—balancing efficiency and risk management. AI can pre-identify customer intent and compile information about conversation history, enabling human agents to understand the customer’s needs more quickly. Still, once the system detects emotional language or sensitive keywords, it immediately hands the case over to a human.

Crescendo Lab provides professional AI potential analysis to help brands identify the most suitable AI services and tools.

Contact Crescendo Lab today to learn about the best AI tools for your brand! 

“After analyzing over ten thousand LINE chat records, we found that 43% of customers’ first text messages were repetitive basic questions. This was seriously consuming our human agents’ valuable time. What we needed was not just a Q&A bot, but an enterprise-grade AI assistant that can precisely route inquiries and support business conversion. Implementing Crescendo Lab’s CAAC AI Agent solution was a key decision to solving our manpower bottleneck and upgrading service quality.”

 

— MR. LIVING | CTO, Lin Yu-Fan

Empowering Human Agents with AI

CAAC AI Agent Builds Intelligent Routing and Proactive Sales

Solution 1: AI Agent Simplifies Through Intelligent Routing, Reducing Time Spent on Repetitive Tasks

After implementing CAAC AI Agent, the system automatically identifies message types and lets AI handle 75% of common questions directly—including store hours, delivery status queries, warranty explanations, and more.

For cases where AI cannot confidently determine the meaning or where messages are highly sensitive, CAAC provides “topic control” and “keyword detection” features to ensure these conversations are immediately assigned to human agents.

This intelligent routing mechanism not only helps MR. LIVING save 160 customer service hours per month, but also allows agents to focus on high-value interactions, improving both service quality and customer satisfaction.

Solution 2: In-House Control of AI Knowledge Base Training – Growing a Brand-Exclusive AI Agent

To make AI the brand’s “second brain,” the first task is to provide it with a large volume of brand materials and customer service knowledge. Early in the evaluation stage, MR. LIVING worried that updating the knowledge base would be inconvenient. Their existing FAQs were managed via CSV files, and they were concerned that updates would have to be sent by email or LINE to technical partners, increasing communication costs and slowing response.

To address this, Crescendo Lab provided a Chatbase-integrated backend, enabling the team to directly upload web links and files themselves, and give instant feedback on AI responses.

For example, when AI once gave an incorrect answer to the question “What sizes does the Puff sofa come in?”, the team could flag the error in the backend and supplement the correct FAQ. The next time, AI answered accurately.

This allowed the knowledge base to continuously evolve and made AI’s answers align better with the brand’s tone and standards—almost like adding a real, continuously learning teammate rather than introducing an uncontrollable tech tool.

Solution 3: A Cross-Department “AI Teacher Team” Turns AI Agent into an All-Round Professional Consultant

MR. LIVING’s expectations for AI Agent went far beyond answering basic entry-level questions—they wanted it to serve deeper and more diverse customer needs. To achieve this, they formed a cross-departmental “AI Teacher Team,” jointly managed by customer service, marketing, product, and engineering:

  • Engineers analyze conversation data to identify high-frequency issues
  • Customer service managers adjust knowledge logic and classification structures
  • Marketing and design teams help test brand tone and overall conversation experience

Through this cross-department collaboration, the brand established a continuous loop of training, testing, and optimization. This not only improved AI accuracy and depth of application, but also reduced incorrect routing to human agents.

The “AI Teacher Team” strategy fully embodies MR. LIVING’s proactive attitude toward embracing AI, demonstrating through action that adopting AI is not just a technical upgrade—it is an organizational transformation.

 

🔑 Key Results 

 

75% – of routine messages answered by AI
81% – accuracy of routing sensitive messages to human agents
160 – working hours saved monthly for customer service

MR. LIVING places strong emphasis on the human element in customer service and is committed to finding a path to grow together with AI. To continuously improve the efficiency and quality of human agents, they have chosen to start with a “Co-pilot collaboration model,” allowing AI to act as a powerful assistant rather than an independent front-line representative—and gradually rolling this out to all store sales teams.

“We hope AI will not replace people, but help people understand customers more quickly, so that customer service can return to the very essence of service.”

 

IMG_5645

— MR. LIVING | CTO, Lin Yu-Fan

This collaboration improves daily efficiency and also opens up a new cycle of “co-learning” between AI and the team: corrections made by customer service become training material for AI, while AI’s suggestions help raise the overall quality of responses.

Uniquely, MR. LIVING’s way of working with AI perfectly reflects the vision that Crescendo Lab aims to realize with its AI Copilot.

AI Copilot Empowers Human Agents – “Suggested Replies + Auto Summary” Optimize Service Flows

Unlike outward-facing AI Agent smart customer service, Crescendo Lab’s CAAC AI Copilot focuses on significantly upgrading the efficiency and quality of internal human agents.

OVERVIEW

Core Functions of AI Copilot

Suggested Replies

Based on the customer’s real-time message and context, AI Copilot automatically suggests several of the most relevant and precise standardized responses or pushes related message templates. Human agents no longer need to spend time typing or searching for information; they can reply with a single click, greatly improving efficiency. 

AI Summaries (Automatic Summarization)

For complex or lengthy conversation histories, AI Copilot can automatically generate a summary and categorize key topics when the conversation ends or before an agent transfers the case. This not only saves agents from time-consuming administrative work but also ensures that colleagues or supervisors taking over can rapidly understand the core issue, delivering a seamless customer experience.

Through AI Copilot, human agents can offload repetitive mechanical tasks to AI and focus their energy on advanced questions that require empathy, professional judgment, and sales consulting—thus maximizing the value of human–AI collaboration.

No.11|AI 自動回覆


💡 AI Copilot provides suggested replies

based on customer needs and integrates conversation records so that human agents can instantly grasp the full context.

AI Agent Levels Up – From Support Assistant to Sales Consultant

As AI performance stabilizes, MR. LIVING plans to transition its AI Agent from a supporting role to a “Senior Sales Consultant” that actively drives revenue. In the future, there will not only be the current AI Support Agent, but also an AI Sales Agent that proactively recommends products and offers, using smart selling to boost conversion rates.

For example, when a customer asks about sofa material, dimensions, or budget, AI can automatically push product cards and comparison information based on the conversation content, creating a “consultative recommendation” where every message has the potential to become a deal.

No. 7|AI 個人化商品推薦

💡 AI Sales Agent actively analyzes

customers’ complex needs and provides integrated comparison and recommended product information.

EXPLORE

Ways the Search Product Feature in AI Copilot Converts Conversations into Sales Opportunities

Seamless Product Search

When customers ask about specific products (e.g., “Do you have electric sofas?” or “I want to see chairs with cat-scratch resistant fabric”), AI Copilot instantly connects to the brand’s product catalog.

Real-Time Product Recommendations and Comparisons

Rather than just replying “yes” or “no,” AI can immediately push product cards, detailed specifications, and even comparison links between multiple items based on customer needs. This removes the friction of customers switching between different sites or apps to search for information.

Drives Pre-Sales Conversion

By shortening product search time to near-instant and providing visual product cards, AI effectively guides customers toward high-value actions like viewing product details, adding to cart, or booking in-store visits—directly boosting pre-sales conversion.

💡 AI Copilot’s Search Product

feature allows human agents to quickly find the right product from a massive information base, saving both time and effort.

Dual Upgrades in Pre-Sales and Post-Sales – Building an Intelligent Online–Offline OMO Loop

With CAAC AI Agent already showing remarkable results during the testing phase, MR. LIVING and Crescendo Lab are shifting their collaboration goal from mere “efficiency optimization” to “business growth and service innovation”—building an intelligent OMO loop across online and offline, and enabling AI to support both pre-sales and post-sales:

  • Pre-Sales Role: AI Agent as Senior Sales Consultant

    AI Agent’s core duty will no longer be only to reply, but to actively participate in the customer journey so that online interactions naturally drive offline visits and conversions. MR. LIVING expects that, in the future, over 40% of incoming inquiries will be handled by an AI Agent, freeing human resources for higher-value sales and relationship-building.

  • Post-Sales Role: AI Copilot for Efficient Internal Collaboration

    In the post-sales phase, AI Copilot is expected to become the brand’s strongest internal assistant—drastically reducing information gaps and redundant work, and enabling the team to focus on cases that truly require human involvement.

“We hope AI will not replace people, but become a second pair of hands—helping the team understand customers faster and focus on areas requiring human judgment and warmth.”

 

— MR. LIVING | CTO, Lin Yu-Fan

Looking ahead, MR. LIVING plans to gradually roll out the CAAC system across its stores, creating three-way collaboration among AI Agents, online customer service, and in-store consultants. This will help them continually balance efficiency and trust, reshape the relationship between service and sales, and turn every conversation into an extension of brand value.

Ultimately, AI will work hand in hand with the brand to realize its mission: “To provide affordable Nordic-style furniture for everyone and enable a better life.”


“You can’t just assign a few customer service staff to execute and assume the project will be successful—that’s impossible. We can’t treat this as something as simple as organizing existing FAQs and service process documents. The knowledge base we give AI is like the textbooks a teacher uses to educate students. We must think about how to build a self-driven teaching team that can continually improve the materials. In this process, we are trying to find a way for 1 + 1 to be greater than 2, by discovering a sustainable, evolving organizational model through collaboration between different teams.”

 

2000px_A7402169 (2)

— MR. LIVING | CTO, Lin Yu-Fan


Contact Crescendo Lab Today for a Consultation!

Crescendo Lab Helps You Build a 360° Customer Experience

  • LINE Gold-Level Technology Partner for 4 consecutive years (2022–2025)
  • Unified access to LINE CRM, Marketing, and Customer Service 360 products
  • Full ecosystem integration with CRM, CDP, and e-commerce platforms
  • 7 billion messages sent per year—trusted by leading brands such as FamilyMart and IKEA