Loading activity…
@hwchase17
@LangChain Always hiring: https://www.langchain.com/careers
AI
A Ramp builder says Laminar took six weeks to build, while its blog post took three months to write.
AI
LangChain says an Academy team member covers how Managed Deep Agents differ from Deep Agents and how to get started.
AI
A user recommends trying Claude Code or Codex before building a complex workflow, saying prompting works surprisingly well.
AI
A post introducing the integration says agents can search one index of Slack, Notion, GitHub and other company content.
AI
A user envisions a continuously updated playbook of rules, decisions and tool guidance, rather than a system holding everything.
AI
A post argues that the harness knows the task, making it the right place for model routing.
AI
A user calls something “the future of coding agents” but does not identify what they mean.
AI
A post weighs whether a finished subagent should interrupt a running parent agent or wait to batch its update.
AI
A user says most of their internal agents have historically acted on behalf of whoever used them.
AI
LangChain says merged-PR rates showed no statistically significant difference from always using its strongest model.
AI
A user shares their Stanford CS146S lecture, with chapters on model changes, agent internals and context engineering.
AI
A post announcing the conversation names a LangChain guest and cites OpenAI's Decisions API and Databricks' ai_decide function.
AI
LangChain says bound tools stay hidden until an agent reads their skill, keeping unused tools out of context.
AI
A poster calls evals the biggest blocker in most agent development pipelines and hints at changes for LangSmith users.
AI
Cognition says Devin updates that memory at night, removing stale records and finding latent information.
AI
LangChain says p0 will give a live booth demo of infrastructure for AI web search, extraction, monitoring and reasoning.
AI
a16z says their podcast also covers Stripe’s acquisition of OpenRouter, AI model costs and dependence on a single lab.
AI
A user describes a Gemini-based system that sends proof drafts to two adversarial verifiers before accepting them.
AI
LangChain says users can publish the apps in a workspace while skipping hosting, authentication and permissions work.
