Today I revisited an old side project — getCert — with the goal of evolving it into a more complete product. Specifically, I wanted to remove the GPG encryption from stored nmap scan data and add PostgreSQL support with a proper schema. Here’s my experience so far.

Firebase is essentially VS Code in the browser, and it handled the initial steps well. I imported my old GitHub repo with no issues, made a few code edits, and was impressed by Gemini’s UI — especially the side-by-side change previews when making suggestions. For basic tasks like editing scripts and drafting the PostgreSQL schema/init script, it worked smoothly.

However, things got trickier when I needed to loop through XML records and insert them into PostgreSQL using psycopg2. Gemini started to fall short here, struggling to generate usable code and understand the context of the task- just began flat out telling me “I’m sorry, but I cannot [do that thing]”.

My only option was to flip over to ChatGPT Canvas. With the same prompts, it quickly generated the function I needed and clearly explained how to integrate it into my main script. While Canvas lacks built-in GitHub integration, it makes up for it with solid IDE-like features and a much deeper grasp of complex code requirements thanks to its larger context window.

I’ve checked the new code into a testing branch — I’ll need to deploy a real PostgreSQL instance to fully test, as neither Firebase nor OpenAI offer hosted DB support directly.

TL;DR:

• ✅ Firebase: Great for quick edits, GitHub integration, and small project tweaks.

• 🚀 ChatGPT Canvas: Ideal for complex logic and generating production-grade code.

It’s encouraging to see how far low-code tools have come — each with strengths that suit different parts of the development cycle.