LangChain Founder: Agents Are Code, Data and Evals Are King
hwchase17 · x · 2026-08-08
Harrison Chase highlighted key perspectives on building self-improving agents while announcing LangChain's new Managed Deep Agents.
The core argument outlines two major principles for scaling agentic systems:
- Agents are code: Since agents excel at writing code, any task can be framed as an agent writing code to define another agent for that task. The optimal engineering loop involves iteratively writing code to define the harness, testing multiple models/configurations, and evaluating.
- Data, Evals, and Environments are king: Teams need zero friction to kick off their data loops. Out-of-the-box tracing should allow agents to constantly mine signals from traces, convert them into evaluations, and drive improvements via harness engineering or post-training.
More from coding & agent
- The Road to Self-Improving Software: Building a Closed-Loop Agent Factory — _AustinCalvert_ · 2026-10-16
- Screenshots show "talk-to-build" AI app tool with one-click open-source publishing to GitHub — yangyi · 2026-10-16
- Multi-agent workflow test: 3 AI agents fact-check each other on a news post — socialwithaayan · 2026-10-16
- Open-Source Tool Turns YouTube Videos into Xiaohongshu Posts with AI — yangyi · 2026-10-16
- GitHub rewrote the 800k-line Copilot runtime in Rust using AI agents, cutting CPU and memory by over 90% — bibryam · 2026-10-16
- What's the hardest part of deploying AI in real businesses that nobody talks about — djgoel123 · 2026-10-16