Recent posts
very helpful
im making a massive "Managed Deep Agents 101" video, diving deep into all the core concepts of managed deep agents: https://docs.langchain.com/langsmith/python/managed-deep-agents-overview would people prefer a single 2hr+ video, or a youtube playlist of like 12 ten minute videos?
LangSmith Engine is hitting the road and the rails in NY + SF If you spot our billboards in-person over the next few months, send it our way!
really excited to see this release AND integrate it with deepagents!! https://www.langchain.com/blog/switchyard-agent-routing-benchmark
Lightning strikes for continuous and long-run agents! Nemotron 3.5 Lightning is smart, fast, efficient and open.
im making a massive "Managed Deep Agents 101" video, diving deep into all the core concepts of managed deep agents: https://docs.langchain.com/langsmith/python/managed-deep-agents-overview would people prefer a single 2hr+ video, or a youtube playlist of like 12 ten minute videos?
Deepagents 👀 We did a bunch of work slimming down the harness, and it shows. Second cheapest harness!
WAIT A MINUE, so Claude is not only the most expensive frontier model, but Claude Code is also the most token-hungry harness? the Claude Code enterprise bills are starting to make sense.
💻How to build a web browsing agent Stagehand v4 is an SDK to let agents browse the web from @browserbase We created a tutorial to show how you could use Stagehand with Managed Deep Agents to create a production ready web browsing agent https://youtu.be/O0hkpChFBkM
Separate brain from hands https://www.anthropic.com/engineering/managed-agents
@hwchase17 Why is the sandbox separate box?
Memory is super interesting and unexplored With managed deepagents, we can be more opioninated about what it should look like. Keep an eye out for what we do here
I think memory becoming its own component is inevitable. The interesting part is what comes right after that: once memory is modular, you need rules for what gets written into it, what expires, what stays local, and what another agent or session is allowed to inherit. Memory becoming infrastructure also turns it into a governance boundary. That feels like the next standardization problem after the component itself.
🎥Managed Deep Agents explained in 20 minutes We launched managed deep agents yesterday. Combines deep agents harness with managed LangSmith infrastructure for a seamless experience Gonna film some more tutorials this weekend - what do you want to see? https://www.youtube.com/watch?v=yi-XZnAVFJg
there's a standard-ish agent stack emerging. theres a bunch of different components, and managed agents solutions package them up nicely
at a really high level, managed agents bundle the harness and infra which makes it much easier to just provide your business context and get an agent!
managed deepagents is one of the launches ive been more excited about in a while wrote a bit about the journey from early langchain to managed agents, and why i think they will be a step change in how easy it is actually run agents
langchain, langgraph, and deepagents are our three core open source projects they each occupy a different place in the ecosystem
own your intelligence if you want an open source starterkit: https://github.com/langchain-ai/open-swe
companies will own their own cloud agent orchestrator
did you know you can do this in langgraph studio? should we make a tutorial for it?
@hwchase17 Replay. Take a failed run, swap one thing, re-run the exact same trace and diff it. Right now debugging an agent means rerunning the whole world and hoping it fails the same way.
we're going to move managed deepagents to public beta this week big focus on all the "boring" and "undifferentiated" infra around agent so you can focus on the agent logic. so far this includes: - opinionated evals setup (using harbor) - memory (agent and user level) - proper oauth for tool access - easy channel integrations (slack, github) - seamless sandbox integration what else would be good?
Internal AI platforms have the potential to transform how companies function Read how Stripe built theirs
"My @Stripe career is divided into before and after Kai." Stripe built Kai, their Knowledge AI Platform, with Deep Agents in one week with just one engineer. Here’s the full story: https://www.langchain.com/blog/how-stripe-built-their-knowledge-ai-platform-on-deep-agents
sharing more about how we evaluate different agents we build internally common themes (from this and future benchmarks): - standardize on harbor - skill for going from traces/raw data to harbor tasks