Sergei·Feb 19Agentic AI: It’s Already Here, You Just Aren’t LookingThese aren’t the droids you’re looking for
Sergei·Feb 11A Thought Experiment: Building an AI Platform Stage to StageArchitecture is not a set of components. Architecture is a history of constraints that a system has accepted and decided not to violate.
Sergei·Feb 8Reflections on Guardrail Architecture: Designing Safe and Scalable LLM SystemsAs large language models become increasingly capable and widely deployed, building a robust guardrail architecture is essential. Guardrails…
Sergei·Feb 7Reflections on Hierarchical Memory and Local Contexts in LLMs with Self-Organizing AGENTS.mdModern LLM agents often operate in complex project structures, where managing knowledge locally can become challenging. The AGENTS.md…
Sergei·Jul 6, 2025Thoughts on External Memory for LLMsFor LLMs external memory could play a role similar to human long-term memory: it helps accumulate experience and apply prior knowledge to…A response icon1A response icon1
Sergei·Jul 2, 20257 Practical Guidelines for Designing AI-Friendly APIsLLMs are not infallible; they are powerful predictors of language, not execution engines.
Sergei·Jun 18, 2025Rethinking AI Agents (practical guide)This post explores AI agents — what they are, what they encompass, and how to foster profitable collaboration between them. Rather than…
Sergei·Jun 15, 202511 Practical Notes about Prompt EngineeringIn this post, I’d like to share a collection of concise yet practical insights I’ve gathered while working hands-on with large language…
Sergei·May 29, 2025Retaining AI Agent Tools Experience with LLM EmbeddingsAI ReAct agents show impressive ability to select tools in order to solve user prompt. But they still have the disadvantage that they can…