Microsoft Ditches OpenAi and Builds Its Own Voice Tech
PLUS: Simple ways to map where AI can help + Quick team readiness check
Start every workday smarter. Spot AI opportunities faster. Become the go-to person on your team for what’s next.
🗓️ September 1, 2025 ⏱️ Read Time: ~4 minutes
👋 Welcome
Big news this week: Microsoft started making its own AI voice technology. But here's what really matters for us in customer experience—the companies winning with AI aren't the ones with the coolest tech. They're the ones solving real problems and measuring real results.
📡 Signal in the Noise
Everything today points to one thing: AI works best when it makes people better at their jobs, not when it tries to replace them. The success stories all have something in common—they started small and built on what already worked.
🎯 Executive Lens
Here's a reality check: A new MIT study found that 95% of companies trying AI projects see no real benefits. But the 5% that succeed? They're not reinventing everything. They're just making their current processes smarter.
Stories That Matter
🤖 Microsoft makes its own voice AI instead of buying it
Microsoft just launched two new AI voice tools that they built themselves instead of using OpenAI's technology. Their new voice AI can create a full minute of natural speech in less than one second. They're already using it in their Copilot assistant for things like daily updates and podcasts. People can even test making custom voices in their labs right now.
Why this matters: When Microsoft builds voice AI instead of buying it, prices usually drop fast. This could make voice automation much cheaper for customer service teams to use.
Try this: Look at everywhere your team uses voice—phone greetings, hold messages, training videos. Add up how much time you spend recording these manually. Voice AI could handle most of this work.
Source: MarkTechPost
🤖 AI starts making big business decisions, not just answering phones
AI isn't just handling customer chats anymore—it's helping executives make major business decisions. Over half of top executives now use AI for their work. LVMH uses AI to watch market trends and make faster decisions. BlackRock has an AI assistant that works all night gathering research for morning meetings. Even Citi is spending billions to let AI help with strategic planning.
Why this matters: If CEOs are using AI for big decisions, customer experience teams should think bigger than just chatbots. AI can help with strategy, not just daily tasks.
Try this: Pick one weekly decision you make that needs lots of data—like staffing levels or which issues to prioritize. Try having AI gather and analyze that information before your next decision.
Source: Forbes
🤖 AI copies of real people might help at work
MIT Technology Review looked at new AI that can act like specific people, not just generic assistants. These AI copies use someone's voice, appearance, and thinking style to handle routine work tasks. They're different from regular ChatGPT because they're trained to think like actual individuals. Early tests show they could fill in for people during busy times, though questions remain about how well they really work.
Why this matters: Instead of one-size-fits-all AI, we might soon have AI that works like your best employees. For CX leaders, this could mean AI trained on your top performers' methods.
Try this: Write down how your best customer service reps make decisions. What questions do they ask? What steps do they follow? This information could train personalized AI helpers.
Source: MIT Technology Review
🤖 Two companies figured out the secret to AI success
Block and GlaxoSmithKline are winning with AI by fitting it into what they already do well, not starting over. Block's engineers use an AI tool called Goose that writes 90% of their code and saves them 10 hours per week. The key? It works within their current setup. GSK uses multiple AI helpers for drug research, but they still rely on human experts to guide the process.
Why this matters: The most successful AI projects don't change everything—they just make current processes better. This approach is easier to implement and more likely to succeed.
Try this: Map out your team's best workflows. Find the repetitive decision points within them. Instead of building new processes around AI, add AI help to these proven methods.
Source: VentureBeat
🤖 Most AI projects fail—here's why
The Wall Street Journal reports on MIT research showing that 95% of business AI projects fail to deliver results, despite companies spending billions. Most organizations saw no benefits from $30-40 billion in AI investments. Workers called custom AI tools "unreliable" and preferred not to use them. The problem? Companies tried to build complicated custom AI instead of adding AI to existing work processes.
Why this matters: This isn't doom and gloom—it's a roadmap for what actually works. The 5% that succeed focus on practical improvements to daily work, not flashy new systems.
Try this: Look at your current AI projects. Are they separate from daily work or built into normal processes? Focus on adding AI to your most important customer service workflows instead of creating standalone AI systems.
Source: Wall Street Journal
💰 Quick hits:
• Microsoft launches first homemade AI voice tech with instant audio creation
• Block says 4,000 engineers save 10 hours weekly using Goose AI helper
• MIT study finds 95% of business AI projects deliver no real value
🎯 Prompt of the Day
Title: Map where AI can help your team
You're helping me understand where AI could improve our customer service. For [YOUR DEPARTMENT], create a simple map showing:
1. **Information we use:** All the places we get customer data (support tickets, surveys, social media, etc.)
2. **Daily decisions:** Key moments where our team has to figure out what to do next
3. **Missing pieces:** Information that would help but is hard to find or access
4. **Repetitive tasks:** Decisions we make over and over that follow similar patterns
5. **Success measures:** How we'd know if better information led to better results
Focus on finding spots where having better information would most help our customers.
This shows you exactly where AI can provide the missing pieces your team needs to make smarter, faster decisions. Use it to find high-impact opportunities that improve decision quality, not just speed.
⚡ Try This Prompt
Help me figure out if our team is ready for AI helpers. Look at: 1) Which of our processes have clear steps and rules, 2) Where we can easily measure success, 3) What customer information is easy to access, and 4) Which team members are comfortable trying new tools. Rank our top 3 best opportunities by impact and how easy they'd be to start.
This quick check helps you find where to begin with AI based on what you can actually do, not what sounds impressive.
💭 CX Note to Self
The best AI projects solve real problems with clear results, not just cool demonstrations.
👋 See You Tomorrow
Today's stories show that AI delivers real value when it's used thoughtfully, while explaining why most attempts fail. The winners measure results and add AI to existing processes instead of rebuilding everything around the technology. What's one current process you could make smarter with AI this week? Hit reply and let me know what you're thinking.
—Mark
💡 P.S. Want more prompts? Grab the FREE 32 Power Prompts That Will Change Your CX Strategy – Forever to start transforming your team, now. 👉 FREE 32 Power Prompts
Special offer for DCX Readers:
The Complete AI Bundle from God of Prompt
Get 10% off your first purchase with Discount code: DI6W6FCD