Handling Context When Building Complex Agentic Systems @ai-science

Uploaded December 2025 | Updated September 2026, 2 weeks ago

Managing context is one of the hardest problems in complex, agentic AI systems. In this video, I share practical techniques that actually work: start with a manual dry run in ChatGPT or Gemini, then ask the model to reverse-engineer the ideal prompt that would have produced the result. Learn why letting LLMs help design better prompts reduces hallucinations, how to transform human-language outputs into agent-ready structured inputs, and how selective context injection (toggling what agents see) improves performance.

#AgenticAI #LLM #ContextManagement #MultiAgentSystems #PromptEngineering #AIEngineering #GenAI #ArtificialIntelligence

Handling Context When Building Complex Agentic Systems

Diving Into Document Question and Answering Systems with LLMs

What is the relationship between language and intelligence?

Before Building an AI Agent, answer these Questions.

The Tension Between AI Governance and Innovation

Strengths, Challenges, and Problem Formulation in RL

What Comes After RAG? The Future of Knowledge Work with LLM Agents

Key Learnings from Building AI Agents: How Open Source Shaped Our Architecture

Relationship between Reasoning and Causality

Using Open Source Framework Versus Industry Standard Like LangChain

Building Software with AI: Why a Simple Task Can Go Wrong

XAI for LLMs: looking under the hood of Large Language Models