I’m building an AI-native backend for developers — would love some brutal feedback

Pawan sri Kumar — Mon, 21 Sep 2026 08:57:21 +0000

I've been building ZendBX, an AI-native backend platform focused on making backend development simpler for developers building SaaS and AI applications.

The idea is to bring the usual backend infrastructure into one place:

PostgreSQL database
Authentication
REST APIs
Row-level security
RBAC
Realtime
Storage
Monitoring & audit logs
API playground
AI-assisted SQL/database tools
MCP support for connecting AI agents and IDEs

The bigger idea we're exploring is:

What if your backend wasn't just infrastructure, but something AI agents could actually work with?

For example, instead of manually switching between an IDE, database dashboard, API tools and documentation, you could connect your IDE/AI agent through MCP and let it interact with your backend within controlled permissions.

We're still building and improving ZendBX, so I'm looking for feedback from developers who actually build with PostgreSQL, Supabase, Firebase, or similar platforms.

What would make you consider switching to a new backend platform?

And what would immediately make you not trust one?

I'd especially appreciate criticism around the product, developer experience, security model, and AI/MCP approach.

ZendBX: zendbx.in

DEV Community: Pawan sri Kumar

I’m building an AI-native backend for developers — would love some brutal feedback