Linear programming: Building smarter AI agents from the fundamentals, part 3

We continue with our series about building agentic AI systems from the ground up and for desired accuracy. In this episode, we explore linear programming and optimization methods that enable reliable decision-making within constraints.

Chapters

  • Introduction and book discussions (0:00)
  • Recap: Utility functions and mechanism design (5:30)
  • Linear solvers versus loss optimizers (8:02)
  • Constraints and optimization problems (12:20)
  • Combinatorial and discrete optimization (17:30)
  • Mixed integer programming for agents (22:32)
  • Episode wrap-up and preview (29:00)

Linear programming and agentic AI: Episode summary

Mathematical optimization might not grab headlines like ChatGPT, but it's the center of AI systems that actually deliver on their promises. In this episode, we explore how linear programming and related optimization techniques provide the essential foundation for building reliable agentic AI.

Starting with the basic concept of utility functions that quantify user preferences, we dive deep into how linear solvers can find optimal solutions within complex constraints. Whether you’re working with constant variables like budget or discrete choices like flight selection, these mathematical techniques make sure your AI respects limits while maximizing value.

We explain common optimization problems like the knapsack problem and shortest path calculations. We show how dynamic programming techniques like branch-and-bound make difficult computational tasks easier to solve. Google Maps is a great example of mixed integer programming in action. It solves both continuous and discrete variables at the same time to find the best routes.

The discussion highlights a crucial contrast between pattern-matching language models and mathematically-grounded optimization. While LLMs excel at generating human-like responses, they often struggle with precise constraint handling. We explore how neurosymbolic approaches might bridge this gap, combining conversational interfaces with the reliability of mathematical solvers.

Ready to build AI that actually respects user constraints? Listen now to understand the mathematical foundations that separate solid, dependable AI agents from those that merely sound impressive but fail to deliver reliable results.

Make sure you check out Part 1: Mechanism design and Part 2: Utility functions. In the next episode, we'll pull all of the components from these three episodes to demonstrate a complete travel agent AI implementation with code examples and governance considerations.

What we're reading

Do you have a question about linear programming and agentic AI?

Connect with them to comment on your favorite topics:

  • LinkedIn - Episode summaries, shares of cited articles, and more.
  • YouTube - Was it something that we said? Good. Share your favorite quotes.
  • Visit our page - see past episodes and submit your feedback! It continues to inspire future episodes.