Your Customers are Already Talking to AI More Than Humans
PLUS: Design Handoffs That Feel Like Warm Introductions
Start every workday smarter. Spot AI opportunities faster. Become the go-to person on your team for what’s next.
📅 October 15, 2025 | ⏱️ 3-min read
🎯 The big picture
Remember when we debated whether AI could handle “real” customer conversations? That debate ended today. These aren’t pilot programs or experiments anymore—these are billion-dollar companies betting their customer relationships on AI that’s so good, customers are choosing it over human agents. If you’re still thinking about AI as a “nice to have,” you’re already behind.
📊 What we’re covering today
• Why Kustomer rebuilt their platform around AI-native architecture
• How Williams-Sonoma’s AI knows your kitchen better than you do (and what that means for personalization)
• The Salesforce-OpenAI integration that makes your CRM conversational
• Microsoft’s vision: humans become the strategic thinkers, AI handles everything else
• Walmart proves conversational commerce isn’t coming—it’s here
1️⃣ Kustomer rebuilt their platform around AI-native architecture
What happened:
• Kustomer scrapped their entire platform architecture and rebuilt it as AI-native from the ground up
• Their new AI Agent Team Assistant doesn’t simply answer questions—it coordinates multiple specialized AI agents for tickets, refunds, and compliance
• Every search, workflow, and customer touchpoint now has embedded AI that learns from context
Why this matters to you: Most CX platforms are still bolting AI onto systems designed for humans. Kustomer realized that’s like putting a jet engine on a horse. They built the jet from scratch. While your current platform is adding “AI features,” Kustomer’s entire platform is AI. This isn’t about having smarter chatbots—it’s about reimagining what customer service software can do when intelligence is baked into every pixel.
The real insight: Your agents spend 60% of their time navigating systems instead of helping customers. AI-native platforms eliminate that friction entirely. The question isn’t whether this approach will win—it’s how fast your current vendor can catch up.
Go deeper: CMSWire
2️⃣ Williams-Sonoma’s AI knows your kitchen better than you do
What happened:
• Williams-Sonoma deployed “Olive,” an AI sous chef that plans entire meals, creates shopping lists, and recommends products based on your cooking history
• Their AI agents now resolve 60% of customer chats autonomously with full access to customer data across all their brands
• Olive doesn’t simply sell—it understands your lifestyle and plans your Thanksgiving dinner better than most people plan their lunch
Why this changes everything: We’ve been obsessed with faster response times and higher CSAT scores, but Williams-Sonoma redefined what customer service means. Olive isn’t answering product questions—it’s becoming a trusted lifestyle advisor. When a customer asks “What should I cook for Thanksgiving,” Olive considers their past purchases, dietary preferences, guest count, and even suggests the right serving dishes.
The CX breakthrough: This is “contextual commerce” at its peak. Instead of treating each interaction as isolated, AI that remembers everything creates relationships that feel personal at scale. Your customers aren’t buying products—they’re buying into an experience that gets better over time.
Go deeper: Chain Store Age
3️⃣ Salesforce and OpenAI made your CRM conversational
What happened:
• Salesforce integrated directly into ChatGPT—you can now query customer records, build reports, and analyze Tableau data by typing questions
• ChatGPT works inside Slack with full context from all your connected tools
• Customers can buy from Agentforce Commerce merchants directly through ChatGPT conversations
Why your team will love this: Remember spending 10 minutes clicking through Salesforce to answer one customer question? Those days ended today. Now you type “show me Sarah’s purchase history and her last three support cases” and get instant answers in plain English. No more training teams on complex interfaces—if they can text, they can work with your CRM.
The bigger picture: We’re witnessing the death of the software interface as we know it. When every enterprise system speaks conversational AI, the barrier between thinking and doing disappears. Your newest hire can access the same insights as your most experienced agent on day one.
Go deeper: Salesforce
4️⃣ Microsoft outlined their vision for agent-operated enterprises
What happened:
• Microsoft unveiled their vision for “agent-operated, human-directed” enterprises where Copilot becomes the universal interface for all business operations
• They’re building contact centers with AI as the foundation, not an add-on to legacy systems
• Microsoft’s own customer support saw significant NPS improvements using this approach
What this means for your career: Microsoft isn’t talking about AI replacing humans—they’re describing a future where humans become strategic directors while AI handles tactical execution. Think less “customer service rep” and more “customer experience conductor.” You’ll spend your time designing better experiences, not executing routine tasks.
The uncomfortable truth: Companies still requiring humans to manually juggle multiple systems will become as obsolete as companies that refused email. The winners will be CX leaders who embrace this “human + agent” model and focus their teams on what humans do best—creativity, empathy, and strategic thinking.
Go deeper: CX Today
5️⃣ Walmart brings conversational commerce to ChatGPT
What happened:
• Walmart integrated with ChatGPT so customers can shop conversationally—no more browsing endless product pages
• Their AI already cuts fashion production time by 18 weeks and customer service resolution time by 40%
• They’re training all associates on AI literacy because the future workforce needs to collaborate with AI, not compete with it
Why this is huge: Shopping became as easy as asking a friend for help. Instead of “search → filter → compare → cart → checkout,” customers say “I need ingredients for healthy weeknight dinners for a family of four” and Walmart handles everything. This isn’t faster—it’s fundamentally more human.
The ripple effect: Every e-commerce site with traditional search and category navigation now looks ancient. When customers can express complex needs in natural language and get perfect results, why would they ever go back to hunting through product grids?
Go deeper: Walmart
⚡ Quick hits
• Anthropic becomes Salesforce’s preferred AI for regulated industries → Claude now works inside Salesforce’s security boundary for healthcare and financial services
• Lila Sciences raises $115M with Nvidia backing → AI-powered scientific discovery hits $1.3B valuation
• Reducto secures $75M for document AI → making sense of enterprise paperwork at scale
💡 CX Prompt Tip of the Day
Design Handoffs That Feel Like Warm Introductions
I need to create seamless AI-to-human escalations that preserve context and momentum instead of forcing customers to start over.
Current situation: Our AI handles [describe main scenarios] but handoffs feel abrupt and customers get frustrated repeating their story.
Design me:
1. Trigger patterns that signal when human help is needed (conversation tone, complexity, emotion)
2. Handoff language that feels like a warm introduction, not AI giving up
3. Context packages that brief human agents perfectly
4. Confidence-building phrases that frame the transition as elevation, not failure
5. Feedback loops so AI learns from successful handoffs
Format your response: Recognition Signals → Handoff Script → Context Transfer → Agent Brief → Success Tracking
Make customers feel they’re being elevated to the right expert, not abandoned by failing technology.
Quick win: Record your next 10 AI handoffs. The best ones sound like AI is introducing customers to a colleague who can help better, not admitting defeat.
🤔 CX reflection
Today’s question: If AI can now run your CRM through casual conversation and plan your customers’ dinner parties, what’s the one uniquely human skill your team should be perfecting to stay indispensable?
(Hit reply—I read every response and often feature insights in future editions)
Talk tomorrow,
Mark