@harikesh_devops Start with just two spans, retrieval and inference, and add the rest later. I believe most teams find the bottleneck in the first week. So which stage was slowest for you
What if AI coding gets so expensive that only companies with funding can afford it?
Think about it.
Weβre building entire workflows around AI agents.
Generating components, writing tests, debugging errors.
Then the pricing changes overnight.
Suddenly your $200/mo bill becomes $800.
Your whole process breaks.
Are we becoming better coders, or are we just renting skill from a model we donβt control?
I donβt have the answer.
But Iβm keeping 2 things in my head now:
> Learn the fundamentals again. So if the API goes away, I can still ship.
> Donβt let AI touch 100% of the workflow. Keep parts that are yours.
Tools should make us faster. Not make us helpless without them.
What part of your workflow would break if AI pricing 3x tomorrow?