Blog

Benchmarks, comparisons, and implementation details.

Fundamentals April 14, 2026 6 min read

Agents Forget. Every Common Fix Trades One Problem for Another.

Four context management strategies on the same task. The one with perfect recall blew the token budget. The cheapest one forgot everything.

Fundamentals Apr 12, 2026 3 min

Agent Memory Depends on a Prompt Nobody Tests

Two summarizer prompts. Same architecture. One recalled 29% of early facts, the other 86%.

LangGraphPromptingContextEngineering

Fundamentals Apr 9, 2026 3 min

Agents Read. They Don't Compute.

The agent fetched the file tree. All 62 Python files were listed. It said 17.

LangGraphEvaluation

Fundamentals Apr 7, 2026 5 min

System Prompts Don't Guarantee Tool Use

Same agent, same 'you MUST' instruction, five different tools. Only three got called.

LangGraphEvaluation

Fundamentals Mar 29, 2026 4 min

Agents Are Easy to Start. Hard to Stop.

Starting an agent is simple. Getting it to stop correctly is where production agents break.

LangGraphReAct

Build Log Mar 22, 2026 1 min

Shipping Beats Planning: Building Across the Entire Agent Stack

The agentic AI space moves fast. Single agents, RAG, multi-agent coordination, voice agents, security, and deep long-running agents. Every agent gets benchmarked.

AgenticAI