- Disruption Now's Newsletter
- Posts
- Six principles of AI and the MidwestCon launch
Six principles of AI and the MidwestCon launch
Rob Richardson
May 28, 2025
In partnership with
Table of Contents
The New Rules of Product Design
Remember when we first discovered the internet? We’d click a blue hyperlink, and like magic, it opened a door to information we didn’t even know existed. That click was simple—but it fundamentally changed how we navigated knowledge. That’s where we are with LLMs today. We’re not just designing new products; we’re building a new way to think, operate, and communicate.
If the internet changed how we find information, large language models (LLMs) are changing how we interact with it—shifting us from navigation to conversation, from retrieval to reasoning, from static tools to dynamic assistants.
But we’re still applying yesterday’s product frameworks to today’s AI problems. That’s why 75% of AI product launches fail. They don’t fail because of the tech—they fail because we’re playing by the wrong rules.
So here’s a breakdown of six principles no one talks about enough.
1️⃣ People need to TRUST it - Every Time
A company named Klarna tried to replace 700 customer service workers with a chatbot. It sounded impressive at first. But customers weren’t happy. They said the bot gave bad answers. Eventually, Klarna had to rehire a bunch of people.
Why did it fail? Because people stopped trusting it.
In AI, trust doesn’t last forever. Every answer the bot gives can make people trust it more—or less. You have to show how the AI came up with its answer, give people the option to talk to a human, and make it easy to undo a mistake.
If people don’t trust your tool, they won’t use it again.
2️⃣ AI Needs a Memory, Just Like We Do
Imagine talking to someone who forgets everything you say five seconds later. That’s how most AI bots work right now.
Klarna’s chatbot didn’t remember past conversations. So customers had to repeat themselves. Again and again. It made the bot feel robotic—and not in a good way.
Real people remember if you were frustrated last time. They remember your name and your last order. AI should do that too.
What you said before
What you were trying to do
How you felt
That way, the next time you talk to it, it’s not starting from scratch.
3️⃣ People Want to Feel Heard—Not Just Get an Answer
Just because an AI can answer fast doesn’t mean it’s doing a good job. If it feels cold, rude, or robotic, people won’t want to use it.
Klarna’s bot gave answers—but it didn’t make people feel like they mattered. That’s why they brought human workers back.
Great AI talks to people with:
A smarter AI remembers:
Respect
Patience
Humility (saying “I’m not sure” when needed)
It’s not just about what the AI says—it’s about how it makes people feel. That’s what builds loyalty.
4️⃣ Build for Mistakes, Not Just Demos
AI is different from regular software. It’s not always predictable. Sometimes it makes stuff up. Sometimes it gets confused.
If you don’t build ways to catch and fix those mistakes, things can go bad fast.
A good AI tool lets people:
Undo anything with one click
See what changed (before and after)
Test things safely before doing them for real
Ask for help from a person if needed
That way, when something goes wrong—and it will—you’re ready.
5️⃣ Show the Win, Right When It Happens
When AI saves people time or catches a mistake, you need to show them. Otherwise, they won’t notice. And if they don’t see, they won’t value it.
Klarna thought they were saving money. However, what they overlooked was the significant loss of trust and customer satisfaction.
A smart AI tool might say:
“This saved you 4 minutes.”
“I stopped 2 duplicate charges.”
“You just kept 3 customers from canceling.”
Show the value clearly, not just in some report months later.
6️⃣ The Second Time Matters More Than the First
Lots of AI tools look cool when you try them once. But what happens the second time? And the third?
That’s what matters.
Klarna’s chatbot had a massive launch. Millions of chats in the first month. Big headlines. But then people stopped using it. Because it didn’t consistently solve problems.
A tool that works once is a demo. A tool that works every time is a product.
You have to design for repeat use, not just first impressions.
My Disruptive Take: We Need New Rules for New Tech
We’re in the early days of AI—just like the early days of the internet. Back then, learning how to click a link changed everything. Now, learning how to work with LLMs is the next big leap.
But here’s the thing: LLMs aren’t just fancy calculators. They’re conversation engines. They’re not built like old software, so we can’t build around them the same way.
The old way of thinking won’t work.
✅ Old thinking: Add AI to your app.
🚫 New reality: Rethink the entire experience with AI at the center.
✅ Old thinking: Build features.
🚫 New reality: Build trust, memory, and emotional connection.
✅ Old thinking: Success = launch day.
🚫 New reality: Success = people come back again and again.
We need a new playbook. One where AI isn’t just a gimmick, but a real partner that understands, adapts, and improves how we work and live.
Let’s build the kind of AI people actually want to use—not just once, but again and again.
Learn about MIDWESTCON 2025
MidwestCon is where algorithms meet empathy, and The Human Code guides the architecture of tomorrow.
This year’s theme, ‘The Human Code: Embedding Empathy into Innovation,’ explores how compassion, ethics, and equity must guide the technologies shaping our world. We’ve designed a series of immersive, hands-on micro-events woven into one powerful experience, equipping real people and real organizations with the tools to thrive in today’s digital landscape while ensuring humanity stays at the heart of every breakthrough.
Keep Disrupting My Friends,
Rob, CEO of Disruption Now & Chief Curator of MidwestCon
Join 400,000+ executives and professionals who trust The AI Report for daily, practical AI updates.
Built for business—not engineers—this newsletter delivers expert prompts, real-world use cases, and decision-ready insights.
No hype. No jargon. Just results.