LangChain releases new features to improve user experience

LangChain has announced a series of new features and updates aimed at improving the user experience on its various platforms, including LangSmith, LangGraph and its generative UI applications, according to LangChain Blog.

LangChain improvements

LangChain introduced the ability to build generative UI applications using LangChain JavaScript/TypeScript, Next.js, and Python. This feature allows users to leverage streaming agent events and tool calls to choose pre-defined components, improving chatbot interactivity. To get users started, detailed tutorials and a Next.js template repository are available.

Additionally, LangChain’s chatbot, LangChain Chat, now allows users to view and continue previous chats, thanks to LangGraph’s backend support.

LangSmith Updates

LangSmith has implemented new workspaces to improve collaboration and organization. These workspaces allow administrators to add users and grant permissions on resources within specific workspaces, simplifying workflows for large enterprises.

LangSmith’s Playground feature has been updated to allow users to start from scratch rather than from a track or prompt. This new tab in the sidebar makes it easier to create and experiment with prompts.

LangSmith also introduced variable mapping for online evaluator prompts, allowing for customizable inputs based on recent runs. Additionally, the platform now supports data retention-based pricing, allowing users to choose shorter data retention periods to save costs.

LangGraph Developments

LangGraph has teamed up with Andrew Ng (DeepLearning) and Rotem Weiss (co-founder of Tavily) to offer a free course on building advanced AI agents. The course covers the implementation of persistence, agent search, and human-in-the-loop functionality.

LangGraph also supports several new integrations, including:

• Meta’s Llama 3 Agents with new codex recipes and video tutorials.
• MistralAI codestral LLM template and completions, which now supports passing a suffix to requests for better results.
• NVIDIA NIM Microservices API for deploying LangChain applications on NVIDIA accelerated infrastructure.
• Nomic Embed Vision for multimodal RAG, enabling embedding and synthesis of images and text.
• Couchbase vector store integration for flexible search capabilities.

LangChain was also recognized by Databricks as a GenAI Partner of the Year and included in their State of AI data report.

Community and learning resources

LangChain continues to support its community with various meetups and learning resources. Upcoming events include:

• June 18: Berkeley LLM meeting in San Francisco.
• June 26: LangChain and Elastic NYC meeting.

For those interested in building practical applications, LangChain offers several tutorials and resources, including:

• A step-by-step guide to building an AI research assistant agent with memory and knowledge management.
• Insights into RAG pipelines to improve search accuracy and relevance.
• Projects on AI-based voice assistants and the latest Whisper models for voice input and transcription.
• Basic tutorial on building LangChain chatbots and managing chat history.

LangChain encourages users to explore these new features and integrations to enhance their AI applications. More details and updates can be found on LangChain Blog and their YouTube channel.

Image source: Shutterstock

Fuente