The Philosophy
Design decisions, architecture thinking, and the principles behind SupaProxy.
AI-native banking
Every software system has three layers. Where humans interact, where rules are enforced, and where work gets done. What happens when AI reshapes all three simultaneously?
Understanding is the highest form of UX
Why we renamed every major navigation concept in our dashboard, and what growing up on the wrong measuring stick taught me about the words we choose.
Knowledge as infrastructure
Most AI platforms treat knowledge as a prompt injection problem. Paste your docs, chunk them, hope the embeddings find something useful. That works for demos. It falls apart the moment you have two teams, three tools, and a compliance requirement.
The pluggable database layer
We extracted every SQL query from core into a separate package. Here is the architecture, the contract, what we gained, what it costs, and where it leads.
Rethinking our marketplace architecture
Why we want teams to build their own guardrails, how we plan to make that safe, and what it means for the platform.
Why we are splitting the server into packages you can swap
Auth, database, queue, vector store. We are turning every layer into a port interface with a default adapter. Here is what we are doing, why, and where we are so far.
Hypothetically, this is how I would build an AI receptionist. What do you think?
Not a chatbot. Not a router. A governed mediator that holds context, enforces policy, and never pretends to know more than it does. This is the architecture we are building.