AppliedAI

Overview
From our AI overlord’s perspective, the world is state machines all the way down. This talk explores how the Finite State Machine and Actor Model—timeless tools for modeling complex, distributed systems—can power Agentic AI workflows to tackle today’s toughest problems. By combining these principles, engineers can build operable systems capable of planning, perceiving, and deciding with clarity, predictability, and control.

We’ll explore how these models apply in the age of LLMs through common Agentic patterns like tool use, reflection, planning, supervision, human-in-the-loop, and a new one I just made up--chartering. All illustrated through the lens of beautiful state charts. Whether you’re on the bleeding edge of AI or seeking proven software design strategies, this talk offers tools to tackle process and system complexity head-on.

Overview

Many of you may remember my previous MinneBar talk "Side Projects for Fun and Profit." Today, I'm sharing a different journey – one that began with a routine checkup revealing concerning biomarkers and led to an unexpected exploration of how AI can transform health management.

Between brief doctor visits every few months, I found myself turning to AI for guidance – asking detailed questions about diet modifications, medication interactions, and lifestyle adjustments. By capturing food images, tracking metrics, and providing health context, I received increasingly personalized guidance that produced measurable improvements in just three months.

This experience highlighted a critical gap in our healthcare system: the lack of intelligent day-to-day guidance between medical appointments. Drawing on my engineering background, I've been developing a solution that combines medical records, wearable data, and AI to provide continuous, personalized health support – turning scattered numbers into actionable insights.

What's particularly promising is how this approach could democratize healthcare for widespread conditions like diabetes, hypertension, and high cholesterol. By making personalized guidance accessible regardless of location or socioeconomic status, we can help address health inequities that disproportionately affect underserved communities.

Join me to explore:

• How AI can decode complex health information and make it personally relevant
• Practical techniques for using AI as a health companion (with live demonstrations)
• The technical challenges of creating memory-aware AI that builds context over time
• Privacy-first approaches to handling sensitive health data

I'll share honest insights from both perspectives – as someone who used AI to improve health outcomes and as an engineer crafting solutions to help others navigate similar paths.

Overview

Over half of all therapeutics approved by the FDA last year came from companies that would be considered startups. Drug discovery as an industry has never been more accessible than it is today, in large part due to AI. We'll discuss in depth the models that made this happen and what to expect in the future.