Mmukundavjcs6inmukundakatta.hashnode.dev·May 15 · 3 min readMost teams running prompt-cached LLM pipelines have no idea what their cache is actually saving them. cachebench tells you.Most teams running prompt-cached LLM pipelines have no idea what their cache is actually saving them. Anthropic shipped the cache-control API months ago. Most pipelines still have not measured the sav00
Mmukundavjcs6inmukundakatta.hashnode.dev·May 15 · 3 min readMost teams find out their RAG pipeline is broken from a complaint. driftvane tells you first.Most teams find out their RAG pipeline is broken from a complaint, not from the data. driftvane is built around the opposite default: the system tells you first. I shipped driftvane on PyPI on May 8, 00
Mmukundavjcs6inmukundakatta.hashnode.dev·May 15 · 5 min readAgent Memory Is Data Infrastructure: A Hermes Plugin That Takes Deletion SeriouslyA few days into using Hermes Agent on my own machine I went looking for the file that holds my conversation history. Hermes ships with a clean MemoryProvider ABC and a list of well-known backends (Mem00
Mmukundavjcs6inmukundakatta.hashnode.dev·May 15 · 6 min read10 Agentic AI Trends Developers Should Watch in 2026Originally published on DEV Community: https://dev.to/mukundakatta/10-agentic-ai-trends-developers-should-watch-in-2026-3ehf Agentic AI has moved from demo theater into a more useful, more awkward pha00
Mmukundavjcs6inmukundakatta.hashnode.dev·May 8 · 4 min readSix Reliability Primitives for LLM AgentsReliability concerns for LLM agents are typically bundled into one heavy framework that asks you to adopt prompting, tool routing, and runtime governance as a single dependency. Production teams want 00
