✨ A New Way of Solving

It's been three years since I first wrote my "Hello World!" program in Python. Honestly, I didn't enjoy that first moment as much as my instructor did, who enthusiastically exclaimed, "Voila!"—as if I'd just performed magic. At the time, I only understood that the program was printing text to the screen, nothing more.

In my first year of programming, it felt like a game of hide and seek—sometimes I was deeply interested, and at other times, I completely ignored it. But as time passed, I started digging deeper. I began to grasp that programming is just the flow of logical thinking—how you reach a solution by applying conditions, looping through tasks, and refining your approach. In essence, programming is a flow, and the more control you have over it, the better programmer you become. More than that, it teaches rationality, logical thinking, and ultimately brings joy. From that moment on, programming truly captivated me, and I never looked back.

Throughout my programming journey, I’ve drawn inspiration from skilled developers who push technological boundaries and create impactful solutions. One of those figures is Jeremy Howard, an AI researcher whose work has profoundly shaped my learning. I first encountered his fast.ai course, Practical Deep Learning for Coders, and found it challenging at first. However, I continued following his work, and when he introduced "A New Chapter for fast.ai: How To Solve It With Code", I saw it as the perfect opportunity to revisit AI. This time, I completed it successfully, gaining valuable skills and tools that have empowered me to build real-world projects and apply AI effectively.

🔥 Leveraging AI the Right Way

In today's fast-paced AI world, problem-solving falls into two extremes:

1. Fully manual coding – A legendary approach that demands deep expertise.
2. Fully AI-generated code – Used by two types of people: seasoned programmers who can verify, debug, and refine AI-generated code (legendary level) and those who rely entirely on AI without understanding what it generates (not-so-cool for someone who enjoys programming).

However, there’s a third category—one that I resonate with: leveraging AI as a tool to assist in problem-solving without relinquishing control. Instead of replacing the programming experience, AI can handle repetitive or tedious tasks, allowing developers to focus on the creative and logical aspects of problem-solving. For example, AI can extract information from text, while programmers focus on how to apply that information to solve a problem. This balanced approach ensures that AI remains an assistant rather than the main driver of the coding process, making solutions more maintainable and bug-free.

This course took this idea even further by introducing a refined concept: Dialog Engineering.

🔍 A Timeless Problem-Solving Strategy Meets AI

Mathematician George Pólya laid out a timeless four-step problem-solving strategy in his book How to Solve It:

1. Understand the Problem – Clearly define the issue at hand.
2. Devise a Plan – Map out your approach.
3. Execute the Plan – Put your strategy into action.
4. Review and Improve – Reflect, refine, and optimize.

Fast.ai is bringing this classic framework into the AI era through Dialog Engineering!

🎓 A New Way to Learn: Dialog Engineering

The "How to Solve It With Code" course introduces a revolutionary concept: Dialog Engineering.

💡 What is Dialog Engineering?

Instead of relying on AI to generate entire code blocks at once, Dialog Engineering promotes an interactive, step-by-step problem-solving approach:

✅ Understand – Break down the problem clearly.

✅ Plan – Decide how to tackle it with AI as a collaborator.

✅ Execute with AI – Write a little code, get AI suggestions, refine and iterate.

✅ Review & Improve – Analyze AI’s contributions, optimize, and refine further.

This method seamlessly integrates Pólya’s principles into AI-assisted coding. Imagine building a LEGO castle with a friend—you place a few blocks, they add more, and together, you construct something remarkable. This dynamic workflow leads to cleaner, more understandable, and more effective code.

🦸 Meet Your AI Coding Sidekick: Solveit

To make Dialog Engineering even more intuitive, Answer.AI has created Solveit, an AI-powered tool designed to guide coders at every step. Solveit follows a step-by-step approach—it doesn’t generate entire blocks of code at once but rather assists in incremental development. With Solveit, you get:

✅ Smart Suggestions – AI-powered guidance tailored to your coding needs.

✅ Context-Aware Assistance – Not just generic code snippets, but meaningful recommendations.

✅ Error Detection & Prevention – Catch bugs early and refine your work efficiently.

By keeping coders in the loop and encouraging iterative problem-solving, Solveit enhances the flow of programming rather than replacing it.

The Answer.AI team has been using Solveit internally to build powerful applications with small teams—now, it’s time to bring that power to the world!

🛠️ My Experience with Solveit and the Course

The course started with solving Advent of Code 2023 problems using Solveit, applying Dialog Engineering while also exploring Pólya’s problem-solving approach. Each lesson introduced useful Python features—like list comprehensions and decorators—while gradually making problem-solving more intuitive.

When Advent of Code 2024 began, I decided to take on the challenge again, and I was thrilled to make it to Day 13! Solveit played a huge role—it allowed me to focus on logical problem-solving while providing valuable AI-powered suggestions. The best part? Solveit follows a step-by-step approach, meaning I never lost control over my thought process, and every step felt like my own achievement.

One exciting aspect of Solveit is the ability to explore frameworks while solving problems. For example, while tackling AoC 2024, I explored NumPy, Matplotlib, and SciPy, deepening my understanding of these libraries.

But Solveit wasn’t just about puzzles—it empowered me to build real projects:

🏗️ My Projects

1. Web Scraping – Before this course, I thought web scraping was complex. A dedicated lesson changed that! I built my first project: 📖 Gutenberg Book Scraper

2. Multimodal AI – After learning about embeddings and Google’s Gemini models, I built: 🖼️ Testing Gemini – A project that generates descriptions based on image prompts.

3. Mathematical Visualizations – Inspired by a YouTube video, I explored fractals: 🔺 Sierpinski Triangle

4. This Blog! – This Blog! – Built using FastHTML and MonsterUI, frameworks introduced in the course that made sharing my projects effortless. Please read this beautiful post MonsterUI- Bringing Beautiful UI to FastHTML by Isaac Flath, Jeremy Howard, & Audrey Roy Greenfeld.

These things I shared are just the tip of the iceberg—there were many other incredible aspects of this course. One standout was the introduction of ShellSage – Your AI Bash Buddy.

This tool integrates AI directly into your terminal, making file handling easier and helping you remember complex commands, ultimately simplifying life for developers.

You can read a fantastic article by Nathon Cooper about ShellSage here: ShellSage - Your AI Bash Buddy.

Another highlight was an inspiring discussion featuring Eric Ries (author of The Lean Startup), Jeremy Howard, and Johno Whitaker, where they explored the impact of modern AI tools, the Lean Startup methodology, and the principle of "build, check, learn" in SolveIt.

Additionally, Hamel Hussain delivered a great talk on blogging—why everyone should write and how sharing knowledge can be empowering.

So there were many more great things in the course. It started with the basics, and by the end, we were learning about cutting-edge technologies like Retrieval-Augmented Generation (RAG). And that’s just the tip of the iceberg! By following an incremental, structured approach—working on small pieces and then connecting them—we were able to build amazing things while maintaining full control over AI-assisted development. Truly an incredible learning experience!

🌟 Shaping the Future of AI-Driven Development

This course has truly transformed how I approach coding. By embracing Dialog Engineering and Pólya’s principles, I’ve gained a deeper appreciation for AI as a collaborator, not a replacement. If you love problem-solving and want to harness AI effectively, I highly recommend "How to Solve It With Code"! 🚀