1. People: Help employees work faster and smarter 

AI should support people, not replace them. Miller says the best AI sits in the background, helping employees do what they already do, only better. 

Example from the keynote: 

Morgan Stanley gave its wealth managers an internal AI assistant. It pulls up relevant documents and answers questions in real time, making the advisor more confident and consistent, while keeping the client experience fully human. 

At her own company, every employee uses a personal AI assistant trained on their team’s goals and values. In meetings, they take “AI breaks”, where the team steps away to generate fresh ideas, then comes back with sharper thinking. 

2. Process: Use AI to streamline repetitive tasks

AI’s biggest wins often come from the back office, where repetitive, manual work slows things down. 

Example from the keynote: 

Walmart piloted an AI negotiation tool called Pactum to handle contracts with smaller vendors—the kind of low-priority deals that often get delayed or overlooked due to limited internal capacity.  

Instead of relying on manual back-and-forth, the AI handled the entire negotiation process autonomously, guiding vendors through terms and mutual interests until a deal was reached.

The results:

• 89% of vendors opted in 
• 68% deal closure (vs. 20% target) 
• 3% average savings per contract 
• 75% of vendors preferred the AI

Miller also shared how she uses AI in hiring.

In one workflow, she gave ChatGPT details about her business and the role she was hiring for. In minutes, it generated job descriptions, KPIs, and even recruitment tweets. Then, she passed it to Claude, which built a working skills test interface—ready to use in interviews. 

3. Product: Improve what you offer—and how you build it 

The last step is using AI to enhance both your product and your production process. 

Example from the keynote: 

Stitch Fix uses GenAI to create 10,000+ product descriptions every 30 minutes. Employees can focus on reviewing and refining, not starting from scratch, cutting turnaround time from two weeks to a few minutes. 

The same AI also powers millions of personalized outfit suggestions each day, tailored to individual users. 

In pharma, one of Miller’s clients uses AI to generate 1,200 landing pages for every new product, each customized to a different market segment. The result? Better targeting, higher performance. 

And with no-code tools like Lovable, Miller showed how it’s now possible to generate full apps from a single prompt. 

1. Multimodality is becoming the default 

AI is no longer just text-based. It’s now fluent across image, voice, video, and even physical interfaces. Expect assistants and browsers that operate seamlessly across formats.  

As Miller shared: 

“I’ll often just go on a 45-minute walk around New York and dictate to Otter AI an entire newsletter or business proposal—and then use that transcript to get started.” 

2. Personal software is on the rise 

Miller believes we’re entering an era where every employee can generate their own tools. Need a packing list? A skills test? A research app? AI can now build “disposable software” that solves a single task and vanishes when it’s done. 

This is how speed scales. Not through big releases, b. But through fast, lightweight solutions tailored to the moment. 

3. Business models will shift to outcomes 

A new wave of SaaS companies are rethinking how they charge for AI. Instead of licenses or seats, Miller predicts we’ll see models based on output: how much value was delivered, time saved, or revenue unlocked. It’s early days, but the shift is coming.