National furniture retailer cuts escalations by 33% and sets a company sales record with agentic AI

The retailer’s original chatbot ran on basic natural language understanding: it matched a customer’s message to the closest known intent and returned a static answer. When a question did not fit a known pattern, the bot had nothing to offer.

• Irrelevant answers drove people to give up on self-service. The prior-generation assistant frequently failed to surface the most relevant knowledge base article, so customers who could not get a useful answer sent the question to a human agent instead.
• Every unresolved chat became a cost. Each escalation meant a live agent had to pick up a conversation from scratch, driving up support costs as order volume grew.
• Pre-sales and post-sales lived in separate silos. The bot could not reliably tell a product question from an order question, let alone route a high-value upsell opportunity to a sales rep versus a routine delivery question to self-service.
• Redundant, friction-filled conversations piled up. Customers often had to rephrase or repeat themselves, and the team needed a more capable system just to keep pace with existing volume, let alone continued growth.

Quiq deployed an agentic AI agent designed to operate across the entire customer journey rather than a single support queue.

• Making the content AI-ready: The retailer’s human agents drew on multiple datasets, including general FAQs, knowledge base content, store location information, and product catalogs. Quiq used large language models to transform this content for AI agent use automatically, without requiring the team to manually reformat anything, and keeps it in sync with the original sources as they change.
• Deep platform integration: The AI agent has read and write access to the systems it needs to act, not just answer. This includes multiple order management and GPS systems for delivery tracking and rescheduling, CRMs with account details and purchase history, the company’s Cordial instance for outbound product recommendation messages, and Google Analytics for reporting. Human agents work inside Quiq’s AI-powered console within Zendesk, and every AI agent conversation writes back to the Zendesk ticket, keeping one unified customer record.
• Agentic AI services for scheduling accuracy: Quiq  consolidates human agent scheduling information scattered across multiple tools and systems, so the AI agent always knows current holiday closures, agent vacations, and business hours before it decides whether to resolve a request itself or route it to a person.
• Orchestration layer: Quiq’s platform combines LLM logic with business rules to identify what a customer actually needs and select the right use case guide, a set of instructions, best practices, and tools, for that situation. This is what lets the AI agent route a high-value upsell to an internal sales rep, or switch a customer from web chat to SMS for delivery updates, without a human designing that path in advance. Pre- and post-answer checks keep every response on-topic and on-brand.
• Agentic reasoning: Rather than following a fixed script, the AI agent evaluates each message against the conversation so far, the current guide, and live information from connected systems, the same inputs a human agent would use, to decide what to do next. That is what allows it to switch topics mid-conversation and handle follow-up questions without losing context.

From a shopper’s perspective, the AI agent notices things a static chatbot never could. A customer browsing the website asks a general question in web chat and gets an answer immediately. When that same customer lingers on a product page afterward, the AI agent proactively reaches out to ask if they have any other questions or would like to talk to a sales rep. The customer says yes, but says they have to go. Rather than losing the lead, the AI agent coordinates a follow-up with a human sales agent over SMS, carrying the full conversation history along with it, so the rep already knows what the customer was looking at and what they asked.

The same proactive, multi-step handling shows up after the sale. The AI agent sends an outbound Apple Messages for Business notification asking a customer to schedule their delivery. The customer books a time, then needs to reschedule. In the same thread, they ask about a required delivery insurance form, a completely different topic. The AI agent handles the reschedule and answers the insurance question without losing track of the original delivery record, coordinating across the systems that hold that information rather than sending the customer back to square one.

For questions outside what the knowledge base can reliably cover, the AI agent is transparent about uncertainty rather than guessing, and offers the option to escalate to a human agent for anything that needs a person’s judgment.

The shift changed the volume of calls that reached a human agent in the first place. Routine order status checks, account questions, and delivery rescheduling requests that once landed in a general queue now get resolved or routed correctly by the AI agent before a person ever sees them. Because every AI agent conversation writes back into Zendesk, agents who do pick up an escalated conversation start with full context instead of asking a customer to explain the issue again.

By replacing a rigid, intent-matching chatbot with an agentic AI agent that can reason across systems, this retailer reduced the volume reaching its human agents while turning chat into a real sales channel.

• 33% reduction in escalations to human agents, freeing the team to focus on complex or high-value conversations instead of routine questions a bot with better reasoning could handle
• 76% self-service support containment rate, meaning the large majority of support conversations now resolve without a human ever picking them up
• 24x increase in Apple Messages for Business interactions, and a 5x increase in SMS interactions, as one-way delivery alerts became two-way conversations customers could actually respond to
• 15% increase in sales chat handoffs, from the AI agent recognizing high-value moments, like a shopper lingering on a product page, and routing them to a human sales rep with full context
• A company sales record, with 10% of total daily sales attributed to chat for the first time, driven by AI agent-led product recommendations
• One platform across pre- and post-sales, replacing a patchwork of static bot responses with a single AI agent that spans product discovery, order status, and delivery scheduling across webchat, SMS, and Apple Messages for Business
• A single customer record, since every AI agent conversation writes back into Zendesk alongside human agent interactions