The evolution of agentic workflows - lessons from coding-assistants

The use of coding assistants has shifted from one-way execution to dynamic back-and-forth. This evolution suggests a universal model for agent integration across industries. Look at the AntiGravity IDE screenshot below, notice the different phases of the workflow in the agent interface:

Building the CollegeROI webapp with a coding assistant - Part 2

I hadn’t planned a part 2 to the original post on creating CollegeROI.app, but I was unhappy with the initial test coverage and quality. Both Gemini 3 Pro and GitHub Copilot struggled to write tests that consistently passed for the application’s core logic.

So round 2 of using coding-assistants to write tests was focused on addressing these issues. In the process, I alse ended up using Google’s new agentic IDE - Antigravity, which has an interesting feature that summarizes the actions taken by the coding assistant, as shown below:

Building the CollegeROI webapp with a coding assistant - Part 1

I built and deployed CollegeROI.app in about 2.5 working days (~20hrs) using coding assistants. This project served two purposes: to experiment with the evolving landscape of AI software development, and to address the increasingly relevant question of college ROI.

First some context

College costs in the US can cause serious sticker shock. To navigate these trade-offs, I built a Google Sheets model to compare options. I found that while tuitions are often similar, starting salaries—and thus true value—can vary significantly.

What was built

I had long planned to convert my Google Sheet into a webapp but lacked the time. However, the rise of AI coding assistants made this the perfect experiment. I chose Google Gemini Pro 3 Preview primarily because it came free with my Pixel phone; otherwise, I would have opted for Claude Code.

Building agents with CrewAI

What was the goal?

The high level goals for this experiment with CrewAI were pretty straightforward:

1. Get some first hand experience with CrewAI by using it to create agents.
2. Understand how crew-ai with the aim to compare it against other frameworks over the next few weeks/months.
3. Learn enough to use it for both work and for side-projects that might arise in the future.

The actual code / agents / tasks were created to write an authoritative and well researched article / blog post on a subject near and dear to my heart - namely Miatas (see post about a future/past Miata-bot here). Could I get a set of agents to write an interesting post about a Miata related topic (namedly after-market Exhaust systems), that I was familiar with? This way, I would not only be in a position to assess the accuracy and quality of what was writte but hopefully learn something new as well.

Exhaust Systems for Mazda Miata (MX‑5): The Ultimate Buyer's Guide for NA, NB, NC & ND

This post is a companion to this one.

Meta description:

Find the best exhaust for your Miata — sound, power, fitment, and legal tips for NA/NB/NC/ND. Brands, dyno expectations, and a practical buying checklist.