Hey Product Hunt! 👋 We built Crowkis because traditional caching layers simply aren't cut out for the probabilistic nature of LLMs. Standard caches rely on exact string matches—if an AI agent tweaks a single word in its prompt, you get a cache miss, racking up massive token bills and slowing down execution. Crowkis changes that. Built from the ground up in Rust for maximum speed, it's completely Redis-compatible and brings semantic intelligence to your data layer. Key features we're excited for you to try: 1. Smart Evaluation: It understands query intent and checks if a cached response is truly safe to reuse. 2. Agent Memory: Seamlessly preserves context and state across complex agentic workflows without repetitive LLM calls. 3. Drop-in Replacement: Since it's Redis-compatible, you can integrate it into your existing stack in minutes. We’d love to hear your thoughts, answer any architectural questions, and learn how you plan to use it for your AI apps!

Crowkis

The smart, Rust-built LLM cache and agent memory layer

The smart, Rust-built LLM cache and agent memory layer