Category
4 posts
← All postsHow to set up Hy-Memory, Tencent Hunyuan's six-layer memory framework for AI agents, with benchmarks, migration gotchas, and lessons from a five-agent fleet.
If you're calling LLM APIs directly from your apps, you're flying blind. A local LLM gateway gives you logs, fallbacks, cost tracking, and a single configuration point for every provider. This post covers why that matters and how to deploy Bifrost in Docker in under five minutes.
Cache hit rate is the single number that determines your LLM bill. A small cache improvement cuts cost in half because cached tokens are 10-20x cheaper than fresh ones. Here's how to get there with Bifrost, OpenRouter session_id, and timing your API calls.
How I gave my AI agents unblockable web access using Firecrawl, Camofox, a residential proxy chain, and SearXNG — all self-hosted on a homelab.