Blog

Ideas on AI infrastructure

Engineering deep-dives, product updates, and lessons from running AI in production at scale.

How we route 1B requests a day across every AI model

A look inside Naagmani's routing engine — how we pick the best-value model in real time, fail over instantly, and keep p95 overhead under 9ms.

How a SaaS customer reduced their monthly AI bill by nearly half without downgrading quality.

Attribute every token and dollar to the customer who generated it — and bill them accurately.

Change your base URL and keep your existing SDK. Here's what to watch for.

Why a single provider outage shouldn't take down your AI features — and how we prevent it.

How security teams can say yes to AI without losing control of spend or data.

The architecture decisions that keep Naagmani's gateway fast at scale.

New articles and product updates. No spam, unsubscribe anytime.

Start building today with a unified API, smart routing, and one dashboard for every model. Free to start — no credit card required.