18/12/2025
# Do you wanna be a player; Understand the game first...
** The AI stack **
* Application:
This is the interface layer. You can put WordPress or any other web or desktop application here.
End user's interactions, input data processing, branding, marketing, etc comes here.
* Orchestration:
All the integrations comes here using protocols like MCP. For example: you want a slack notification or do something using a external tool.
* Data & RAG:
AI models have limitations. Most of them are trained using public generic data. They require knowledge. If you want them to do something specific, then you also have to provide relevant context. Models also can't process big datasets, so you need to break data into small chunks and retrive most useful data for the Model.
..
"Above three stacks are the playground for most of tech companies and most of the software engineers."
* Models:
The mighty AI models. This is the playground for giant tech companies, because it requires huge amount of resources and money. Less then 0.5% software engineers/researchers around the world, are able to create a production ready Model from scratch - this is true for now and also for near future.
* Infrastructure:
This is a evergreen playground. Everything in the digital world require infra. Whatever you do using AI, they have to be hosted in some sort of infrastructure. And most of the cases it's gonna be cloud infra. So big techs are the providers, small tech companies are the consumers and DevOps engineers are Messi-Ronaldo of the entire ecosystem. Because AI can't replace them, at least within nex five years.
