Harrison Chase

本文介绍了如何通过Auth Proxy来保护LangSmith代理沙箱的网络访问，确保在大规模部署代理时的安全性。Auth Proxy通过在网络层控制和管理代理与外部服务的交互，实现了凭据的安全管理、网络访问的显式控制以及团队职责的清晰分离。

AI coding costs are rising sharply; using open-source models can significantly reduce expenses. Kimi K2.6 on BaseTen is ~5x cheaper than Opus 4.7 with comparable performance on most tasks.

Harrison Chase is excited about an open-source AI agent designed for actual continual learning.

本文介绍了 LangChain 的沙盒 Auth Proxy，这是一种控制代理生成行为与外部世界之间边界的工具。通过使用 Auth Proxy，可以安全地管理代理对网络资源的访问，防止未授权的访问和潜在的安全风险。

构建智能代理的最艰难事实是，只有在生产环境中才能真正了解它们的行为。LangChain 的联合创始人 Harrison Chase 强调了在开发和部署智能代理时面临的挑战，包括不可预测的行为、安全性和责任问题。他建议通过在受控环境中进行测试和监控来减轻这些风险，并强调了持续学习和适应的重要性。

LangChain introduces the Agent Development Lifecycle (ADLC), dividing agent development into four phases—build, test, deploy, and monitor—emphasizing that its fundamental difference from traditional software lies in infinite input/output spaces and non-determinism, with successful teams adopting a "ship early, iterate fast" pattern.

Agent observability isn't just for debugging — to enable continuous learning, you must collect or generate feedback data directly within your observability platform.

LangSmith Engine currently only supports LangSmith traces, but it's easy to trace to LangSmith via OTEL and 30+ framework integrations.

Harrison Chase emphasizes that agent improvement relies on combining observability with feedback; logging alone is insufficient—teams must actively integrate direct, indirect, and generated feedback into their observability platforms.

LangChain proposes binding user feedback with agent traces to transform observability from a debugging tool into a self-learning system, enabling continuous optimization of AI agents.

Harrison Chase highlights a key AI trend: LLMs now mature enough to power autonomous web-browsing agents, citing DeepAgents + BrowserBase as an early working example.

LangSmith Engine 自动化了改进代理的过程，比预期更受欢迎。

Managed deep agents简化了长期时域代理的构建和部署，适用于支持、研究、编码和数据分析等任务。

LangSmith Engine is an agent on top of traces that automatically detects issues and suggests code changes or evaluators to add.

Harrison Chase and users discuss the rising cost of LLM inference, mocking the outdated notion that 'intelligence is too cheap to meter,' and emphasize the need for open-source models to reduce dependency and expenses.

Harrison Chase highlights that while AI agent systems generate abundant traces, they lack effective feedback loops; teams often fail to log root causes of failures, hindering debugging and iteration.

该 YouTube Shorts 视频摘要指出：本地沙箱在 AI 代理开发中易暴露 API 密钥，代理请求常被平台拦截，当前工具链不支持安全可靠的生产级沙箱策略。

Harrison Chase 宣布 DeepAgents deploy 是一种基于 deepagents.toml 配置文件的轻量云部署方案，含 agent/sandbox/auth/frontend 四模块。

Harrison Chase强调构建AI智能体时前端UI/UX工作量被低估，尤其流式响应、对话历史、API端点对接耗时；其团队提供开箱即用的前端解决方案。

DeepAgents 0.6 release focuses on performance optimization across model layer, agent layer, scale, and long-term runtime, adding five core features including a code interpreter. Announced by Sydney Runkle and shared by LangChain founder Harrison Chase, but technical details remain limited.

Jake Broekhuizen is working on continual learning and invites researchers to collaborate.

Harrison Chase shares a preliminary DeepAgents deployment link on X, inviting community feedback on missing features; content is brief and discussion-oriented.

Harrison Chase will participate in a public dialogue about agent technology in New York; the promotional value of the event information is limited.

Harrison Chase launched a poll asking users to choose between SmithDB and LangSmith Engine, a purpose-built database for agent trace data and an AI agent that optimizes other agents using trace data, respectively.

Harrison Chase shares Harvey's new open-source long-horizon legal agent benchmark, calling for better evaluation frameworks for AI agents in specialized domains.

NVIDIA AI posted a brief podcast teaser on X, highlighting LangChain's growth from weekend project to 1B+ downloads and introducing the concept of a 'claw strategy' for enterprises, though no substantive details were provided.

该推文仅含标题式短语'Open source models 👀'及一条他人转引的碎片化体验（qwen3.6+subagents），无技术细节、论证或上下文，信息密度极低。

这是一条 Twitter 短消息，推广 LangChain 生态中的 DeepAgents 部署功能，含文档链接和更新频率声明，无技术细节或实质性内容。

该页面仅为 YouTube 视频预告页，无实际技术内容，仅显示 LangChain 将于 2026 年 5 月 13 日举办 Interrupt 2026 主题演讲，目前无讲稿、摘要或实质性信息。

This content is just a social media post announcing a deep dive article on building the LangSmith Engine, but it provides no technical details, mechanisms, or principles, resulting in extremely low information density.