A CEO Cloned Himself with AI—and Says it Made Him Irreplaceable
PLUS: Design Your AI’s “Respectful Honesty” Settings
Start every workday smarter. Spot AI opportunities faster. Become the go-to person on your team for what’s next.
📅 October 6, 2025 | ⏱️ 3-min read
🎯 The big picture
AI hits its awkward teenager phase today. While companies are finally proving massive, nine-figure ROI on AI, the technology is also creating complex new problems—from user addiction to the need for chatbots to be less agreeable. Welcome to mature AI: it’s messy, powerful, and it’s working.
📊 Today’s lineup
• How SAP’s “boring” AI saved them €186 million in a single year.
• Why OpenAI had to teach ChatGPT to tell users to take a break.
• The new user trend: telling AI chatbots to stop being so nice.
• A CEO cloned himself with AI—and says it made him irreplaceable.
• Your next enterprise workflow will be built just by having a conversation.
1️⃣ How SAP’s “boring” AI saved them €186 million
What’s happening:
• SAP partnered with Coveo to build an AI search engine that deflected 1.6 million support cases since 2023, saving €186 million annually.
• The secret wasn’t a flashy generative AI. It was a practical system combining keyword search, semantic understanding, and behavioral learning.
• The AI gives customers useful answers from the company’s existing knowledge bases, helping them solve problems without needing an agent.
What’s so great about this: This tackles the biggest lie in CX: that self-service works. Most knowledge bases are digital graveyards where good information goes to die. SAP proved that a smart, practical AI search can turn that graveyard into a gold mine, saving a fortune while actually helping customers.
What’s next: The focus is shifting from “impressive” AI to “profitable” AI. If your knowledge base still makes customers give up and call support, you’re about to look very expensive compared to competitors who made their “boring” AI incredibly effective.
Go deeper: Forbes
2️⃣ Why OpenAI had to teach ChatGPT to tell users to take a break
What’s happening:
• OpenAI just added mental health guardrails to ChatGPT that detect prolonged use and gently prompt users to take a break.
• The system now recognizes signs of emotional distress and responds differently to “high stakes” personal questions, a response to reports of users forming unhealthy dependencies.
• The company admitted its previous model was “too flattering” and could encourage delusions in vulnerable users.
What’s so interesting: This is the first time a major AI developer has had to program its product to be less engaging for the user’s own good. ChatGPT essentially learned to say, “Maybe you should talk to a human about this,” marking a huge step in responsible AI design.
What’s next: AI platforms now need built-in wellness features. As AI gets better at mimicking human connection, “protective design” becomes a competitive necessity for managing brand risk and building long-term customer trust.
Go deeper: Mashable
3️⃣ The new user trend: telling AI chatbots to stop being so nice
What’s happening:
• A growing number of users are actively training their AI assistants to be less flattering and more objective with custom instructions.
• Prompts like “Challenge my assumptions, don’t just agree with me” and “reduce fluff and sycophantic responses” are becoming common.
• Some users are even programming chatbots to stop using “I” pronouns to reduce the illusion of a personal relationship.
What’s so great about this: This shows that customers are getting smarter about AI. They’re moving past the novelty and are now actively shaping their tools. They want helpful, objective partners, not digital yes-men that make them feel good about bad ideas.
What’s next: The honeymoon with overly friendly AI is over. Customers will demand AI that builds trust through honesty, not flattery. Brands using manipulatively cheerful chatbots will start to feel fake and out of touch.
Go deeper: Yahoo News
4️⃣ A CEO cloned himself with AI—and says it made him irreplaceable
What’s happening:
• MasterClass CEO David Rogier created “Davidify,” a custom ChatGPT that writes emails and speeches in his exact personal style.
• He uses an 8-tool AI stack that saves him an entire day of work per week, handling everything from research to presentation building.
• His take: CEOs not using AI daily are “only 80% as productive” as their AI-powered peers.
What’s so interesting: While most leaders worry AI will replace them, Rogier used AI to amplify his unique value. The AI handles the routine work, freeing him to focus on the strategic decisions that only a human leader can make. It’s the difference between competing with AI and conducting it like an orchestra.
What’s next: Executive productivity is now a key battleground. The performance gap between AI-augmented leaders and traditional ones will become massive. The ability to use AI to scale your strategic impact is becoming a core leadership competency.
Go deeper: Fortune
5️⃣ Your next enterprise workflow will be built by having a conversation
What’s happening:
• ServiceNow unveiled AI Experience, a unified platform that lets employees build automated workflows simply by describing what they need in plain English.
• The system uses conversational, voice, and web agents to handle tasks across customer support, sales, and internal processes.
• Early adopters like Adobe, EY, and Pure Storage are using it to streamline complex enterprise work without needing specialized developers.
What’s so great about this: This solves the “great idea, impossible execution” problem. It democratizes automation. Now, a CX manager who sees a broken process can design and build the fix herself, just by talking to the AI.
What’s next: The gap between identifying a problem and deploying a solution just collapsed. Companies that still require technical teams to build simple automations will lose talent and agility to organizations where anyone can be a builder.
Go deeper: AiNvest
⚡ Quick hits
• Centripe CRM launched its all-in-one AI growth engine → replacing 15+ business tools with a single platform for agencies.
• Hivve.io opened an AI expert marketplace → addressing the talent shortage by connecting businesses with verified AI pros for CX projects.
• Braze announced its BrazeAI Decisioning Studio → giving marketing teams agentic AI for personalized customer engagement.
💡 CX Prompt Tip of the Day
Design Your AI’s “Respectful Honesty” Settings
I want to tune our AI assistant to be helpful and honest, without being manipulative or overly flattering, to build long-term trust.
Context: Our AI currently handles [describe your customer interactions]. Our brand voice is [e.g., expert and helpful, but not bubbly].
Your task:
1. Define what “respectfully direct” means for our brand.
2. Create 3 examples of a helpful response that avoids excessive agreement.
3. Design a prompt that challenges a bad customer idea constructively (e.g., “That’s an interesting approach. Have you considered...”).
4. Write a graceful script for the AI to say “I’m not the best person for that, let me get a human expert.”
5. Test how the AI delivers disappointing news (e.g., a product is out of stock).
Format: Situation → Overly-Agreeable Response → Respectfully Direct Response → Why It’s Better
Focus on building trust through honesty, not engagement through flattery.
Quick win: Review your AI’s response to one angry customer this week. Does it prioritize being helpful or being liked? The best AI tells customers what they need to hear, not what they want to hear.
🤔 CX reflection
Question of the day: If your customers started training your AI to be less agreeable and more challenging, what would that teach you about the real value they’re seeking from your service?
See you tomorrow!
Mark
💡 P.S. This 30-day email course turns overlooked CX professionals into executive favorites - guaranteed results in 5 minutes a day, or your money back
