This one was different
I’ve built AI systems before.
RAG pipelines. Automation tools. Dashboards.
But building RiskLens CI?
This one pushed me in a different way.
Not because the idea was hard…
But because the environment was.
The constant back and forth
This wasn’t a straight build.
It was:
- switching between tools
- fighting with extensions
- reconfiguring environments
- debugging things that should have worked
And doing it over and over again.
One minute everything runs.
Next minute something breaks for no clear reason.
The hidden complexity of “AI-powered dev”
From the outside, it looks simple:
“Just use AI, build fast, ship.”
But in reality?
You’re dealing with:
- API inconsistencies
- environment mismatches
- dependency conflicts
- version issues across tools
- cloud vs local differences
And when you’re building something like a CI assistant…
Everything has to work together.
The part nobody talks about
The frustration.
The restarting.
The feeling of:
“Why is this breaking again?”
It’s not just technical.
It’s mental.
Because you’re not solving one problem…
You’re solving layers of problems at the same time.
What RiskLens CI is teaching me
This project is teaching me:
- how to debug across systems
- how to work through unstable environments
- how to stay consistent even when progress feels slow
- how to build something that actually works
Real engineering vs perfect demos
There’s a difference between:
- a demo that works once
- a system that works consistently
RiskLens CI is forcing me into the second category.
Final thought
If you’re building with AI right now and feeling frustrated…
You’re not doing it wrong.
You’re just in the part nobody posts about.
👉 Full blog here:
https://bdcreativesystems-star.github.io/ai/career/machine%20learning/2026/03/19/getting-certified-in-ai-ml-while-watching-vibe-coders-move-faster.html
Top comments (1)
Curious—what’s been the most frustrating part of building with AI for you?