Search: “langchain-blog”

4 results

Each lane shows up to 8 recent matching entries, ordered from earlier to later. Tracks load separately to keep the 75,000+ entry wiki fast.

Tools

ToolCursor1 recent entries

9 Apr 2026Human judgment in the agent improvement loop

AI agents work best when they reflect the knowledge and judgment your team has built over time. Some of that is institutional knowledge that’s already documented and easy for an agent to use as-is. Bu

ToolLangChain4 recent entries

8 Apr 2026Better Harness: A Recipe for Harness Hill-Climbing with Evals

By Vivek Trivedy, Product Manager💡TL;DR: We can build better agents by building better harnesses. But to autonomously build a “better” harness, we need a strong learning signal to “hill-climb” on. We