Why I'm Betting on Rust and AI
My Journey from Experiments to Production AI agents
Like many developers, I’ve spent years chasing the next big thing, Flutter, RN, React apps, and web3. But lately, I’ve found myself at an intersection that feels different. More substantial, more inevitable: Rust and AI.
It wasn’t a sudden epiphany. It was more like watching two parallel trends converge until ignoring them felt foolish. On one side, AI is exploding beyond chatbots and demos into real production systems that need to interact with databases, APIs, and business logic. On the other hand, Rust is quietly maturing from the “interesting but niche” language into something companies actually bet their infrastructure on.
The moment it clicked for me was realizing that AI’s biggest bottleneck isn’t intelligence anymore; it’s integration. These models are incredibly smart, but they’re isolated. They need to reliably, safely, and quickly talk to the real world. They need the exact qualities Rust was designed to provide.
I’d seen too many promising AI projects stumble in production. Memory leaks in long-running agents. Race conditions in concurrent tool execution. Mysterious crashes when handling unexpected data. The kinds of bugs that are “acceptable” in a prototype but catastrophic in a system, making real decisions.
That’s why I decided to go all-in on Rust for AI infrastructure. Not because it’s trendy (though it is), and not because it’s fun to write (though it is). But because building AI systems that actually work in production requires the kind of guarantees that only Rust provides.
This isn’t about replacing Python for model training or experimentation; that ship has sailed, and Python won. This is about everything that happens around the model: the servers, the integrations, the tools, and the infrastructure that makes AI useful rather than just impressive.
This is also why I’m building a Code Review Agent Instalog AI with Rust/TS and Railway. I’m pushing to see how far I can push Rust as a convenient tool to build fast, reliable, and memory-safe agents.
So expect more content from me in this regard. Rust, TS, and AI. That’s where I’m at these days.



Good write up!
This bit of the write up that states why rust:
——————————————————————-
Memory leaks in long-running agents.
Race conditions in concurrent tool execution.
Mysterious crashes when handling unexpected data. The kinds of bugs that are “acceptable” in a prototype but catastrophic in a system, making real decisions.
——————————————————————
Rust doesn’t guarantee you won’t run into any of these. I think speed (no garbage collection) and guaranteeing there’s no memory corruption (use after free kind of bugs) are good reasons to choose rust.
Also, if many of the integrations are happening against C / CPython libraries, this will mean a lot of unsafe code without any rust guarantees.