Denis Urayev·Mar 22Self-host LLM Won’t Save Data Privacy: Why Companies Still Move to the CloudThere is a popular idea:A response icon1A response icon1
Denis Urayev·Jan 13Why the MCP Standard Might Quietly Fade Away?Over the past year (2025), the Model Context Protocol (MCP) has become one of the most discussed standards for connecting LLMs with the…A response icon3A response icon3
Denis Urayev·Nov 18, 2025RAG Is Dead? Before You Build It, Try File-First AI AgentI replaced a complex RAG setup with a file-first agent. Simpler, faster, more accurate. You might not need a LangGraph pipeline at all.A response icon26A response icon26
Denis Urayev·Oct 10, 2025One prompt engineering technique that might save you next timeNot long ago, my colleague asked me to help with what seemed like a straightforward task which GPT-4o can’t solve. We had three tables with…A response icon1A response icon1
Denis Urayev·Oct 10, 2025The AI Cost Illusion: Why Cheaper Tokens Don’t Mean Cheaper AINot long ago, I shared techniques for compressing large Excel files so they could be sent more efficiently to LLMs.A response icon1A response icon1
Denis Urayev·Aug 5, 2025A New Era in Software Architecture: From Reusable to Regeneratable CodeFor decades, reusable code has been the holy grail of software engineering. We’ve been trained to build abstractions, follow SOLID…A response icon1A response icon1
Denis Urayev·Aug 5, 2025How to Fit Massive Excel Files into LLMs: The Spreadsheet Compression PlaybookTabular data is the lifeblood of virtually every organization. From sales reports and financial ledgers to inventory lists and customer…