Insights & Ideas

Latest blog posts

Discover stories, tips, and resources to inspire your next big idea.

Research

With Anyray, Claude solves more tasks, faster, at 32% lower cost

Gidi SteinbergYarden Shechter
Research

Why model routing backfires in agent sessions, and what a switch really costs

Yarden Shechter
Research

The decision engine: how Anyray learns what is safe to cut

Gidi Steinberg
Research

Why context compression breaks prompt caching, and how to keep the prefix stable

Dean RubinYarden Shechter
Research

How retrieval handles make lossy compression safe

Yarden Shechter
Engineering

cacheLint: detect prompt-cache breaks in LLM agents

Yarden Shechter
Engineering

Running Claude Code behind an AI gateway: the 200k context trap and the env vars that fix it

Dima Rudenko
Engineering

Fail-open by design: how Anyray keeps your traffic flowing when something breaks

Anyray team
Engineering

How Anyray works: cutting tokens before the model without breaking the cache

Dean Rubin