Welcome to June, where things at LangChain are heating up! In this edition, learn about organizational enhancements to LangSmith plus improvements to the playground and online evaluator prompts. Our LLM-generated UI is also generating buzz, and we’ve cooked up some delectable code recipes using Llama 3.

And don’t sleep on the learning resources, from a free LangGraph course with Andrew Ng (DeepLearning) to community insights on your favorite topics like RAG and agents.

Product Updates

Highlighting the latest product updates and news for LangChain, LangSmith, and LangGraph

LangChain

🧬 Build generative UI applications using LangChain JavaScript/TypeScript, Next.js, and Python

Using streaming agent events and tool calls to pick pre-built components, you can now use generative UI to improve your chatbot with interactable components. Get all the information you need in a single UI.

Learn more from our 3-part generative UI video series on YouTube, covering JavaScript and Python, and check out our Next.js template repo for examples.

💬 Chat LangChain improvements

Users can now view and continue previous chats in Chat LangChain, our chatbot for LangChain Python docs and API reference, thanks to LangGraph running under the hood.

LangSmith

LangSmith now offers Workspaces to separate resources for different teams or different environments. Admins can add users to Workspaces, granting them permissions only on resources (projects, prompts, datasets, etc.) within those Workspaces.

Read our blog post to see how Workspaces can streamline workflows for large enterprises. For finer-grain access control, reach out for the Enterprise plan.”

The Playground is now its own tab in the sidebar of LangSmith. To create a new prompt, simply craft a prompt in the empty playground and click "Save As" to name it. This streamlines prompt creation and allows for playground experimentation with a straightforward entry point. Try it out here.

In LangSmith, you can now use any structured prompt from the LangChain Hub with customizable variables. You can personalize your inputs based on recent runs to match your schema.

💳 Updates to LangSmith Pricing

As of this month, LangSmith now supports data retention based pricing. Choose a shorter data retention period on your traces for savings. Explore the docs or our tutorial on optimizing spend for more info.

LangGraph

DeepLearning Course on AI Agents in LangGraph

We've teamed up with Andrew Ng (DeepLearning) and Rotem Weiss (Tavily co-founder) to teach a free course. You’ll learn how to build advanced agents using LangGraph — our framework that helps you balance control and autonomy for agentic workflow.

Topics include implementing persistence, using agentic search, and incorporating human-in-the-loop. Sign up now for free.

Upcoming Events

Meet up with LangChain enthusiasts, employees, and eager AI app builders at the following IRL events this coming month:

🐻 June 18 (San Francisco): Berkeley LLM meetup. Calling all Berkeley PhD students, faculty, and alumni! Come learn about LangGraph, cool OSS projects built at UC Berkeley, and connect with likeminded peers working on LLMs. Sign up here.

🏙️ June 26 (NYC): LangChain and Elastic NYC meetup. Hear lightning talks, meet some of the LangChain team, followed by networking, pizza, and refreshments. Sign up here.

Collaboration & Integrations

We 💚 helping users leverage partner features in the ecosystem by using our integrations

Open source / local agents with Meta Llama 3

Check out our new recipes and video for Meta’s Llama 3 agents using LangGraph. Learn how to build LangGraph tool-calling agents, RAG agents for self-corrective control flow, and how to run RAG agents locally with Nomic and Ollama.

Mistral’s codestral model and completions LLM supported in JavaScript

LangChain supports MistralAI’s latest codestral and completions model, which now allows for passing a suffix to prompts — which can improve results for “fill-in-the-blank” coding.

Integration with NVIDIA NIM

Use LangChain with NVIDIA’s NIM microservices API to deploy anywhere on a single command on NVIDIA accelerated infrastructure. See an example of how to build a RAG agent with LangChain and NVIDIA NIM.

Nomic Embed Vision support for multimodal RAG

Embed images and text with Nomic Embed Vision and Text, then retrieve with similarity search and synthesize answers with LangChain for a multimodal LLM.

Couchbase vector store integration

LangChain supports vector search in Couchbase to pull in relevant queries and enable flexible search capabilities.

And a special thank you to Databricks for naming LangChain as their GenAI Partner of the Year and including us in their State of AI Data Report 🙌

Speak the Lang

See how our 1M+ developers and builders are using LangChain, LangSmith, or LangGraph in their day-to-day. Thank you for always helping us build better!

Build and execute with agents

Agents are all the rage for task execution, but can have limitations. Eden Marco (LLM Specialist @ Google) explains why autonomous agents can fall short, and how having controllable agents (like in LangGraph) can overcome these issues.

For a practical guide to agents, look to this step-by-step tutorial from Richmond Alake (Staff Developer Advocate @ MongoDB) on building an AI research assistant agent with memory and knowledge management.

Other notable agent projects:

• Tribe: Low code tool to rapidly build and coordinate multi-agent teams by Jerron Lim (SWE @ LTA Singapore)
• Generative UI financial agent with LangChain agents and Vercel by Virat Singh (SWE @ Faire)

RAG: From POC to Production

RAG pipelines can enhance the accuracy and relevance of search results — which is exactly what Jeff Nelson (Developer Advocate) and Ashley Xu (SWE) from Google help you do via simple RAG on BigQuery with LangChain.

To see how to take your RAG app from prototype to production, watch Hassan El Mghari’s (DevRel @ Together AI) PDF to Chat project — which also highlights trace visibility using LangSmith (see 40:35 in video).

Additional RAG use cases to consider:

• Building RAG with PowerPoint presentations by Maria Khalusova (Dev Advocate @ Unstructured)
• IncarnaMind: Notion-like notebook to chat with personal documents by Junru Xiong (Data Scientist @ R2 Factory)

Speech-to-text, Text-to-speech

If you’ve ever thought about creating an AI-powered voice assistant, Karim Lalani (SWE) has you covered with an AI Voice Chat built on WebRTC and LangChain.

Or, if you’ve wanted to try the latest Whisper models, explore this OSS AI Devices project supporting voice input, transcription, text-to-speech, and more with conditionally-rendered UI components.

Back to Basics

And of course, we can’t forget the basics. Learn how to build the most basic LangChain chatbot from Cobus Greyling (Chief Evangelist @ Kore AI) if you’re new to LangChain.

And for those a little further in, you can manage the chat history of your LangChain apps with Dragonfly's context management or build your first agentic graph by following the LangGraph essentials video.

Other useful projects:

• Website Q&A RAG Chatbot with Replit, LangChain, and OpenAI that answers questions on docs and provides blog summaries
• Building DevOps AI Assistant with LangChain, Ollama, and PostgresSQL by Huseyin Babal (Cloud Engineer @ Namecheap)

How can you follow along with the Lang Latest? Check out the LangChain blog and YouTube channel for even more product and content updates. For any additional question, email us at support@langchain.dev.