• Unpacking LangStream's Event-Driven Architecture

    Customers expect instant responses. So, modern applications demand real-time capabilities as they process user input, update analytics, and adjust to rapidly changing situations. Event-driven architectures can accommodate this demand by handling real-time data as it emerges.

  • LangStream 0.4 Features

    If you were to ask 10 people what “streaming” is, you would get 10 different answers. The only thing they would agree on is the timing… streaming is processing data in real-time. The minute the data is written to a persistent store and then queried, you’ve lost real-time.

  • Using Ollama in LangStream

    Ollama is a tool that allows you to easily deploy and manage Open Source LLMs. You can even run the Large Language Model locally and augment it using the Retrieval Augmented Generation (RAG) technique.

  • Building Scalable Vectorization Pipelines in LangStream

    The most fundamental component of a retrieval augmented generation (RAG)-based application is your data. You store your data in a vector database as documents and query it to retrieve the most relevant documents for a given input text.

  • Building a Q&A RAG chatbot with conversational awareness

    You want to build a chatbot that answers questions about private data. For example, your chatbot taps into your company’s documentation and other knowledge sources to help support agents answer questions from your customers. Or maybe it’s a support chatbot to directly answer questions from your users.

  • Re-Ranking with Maximal Marginal Relevance in LangStream

    When implementing the retrieval augmented generation (RAG) pattern, it’s necessary to query a vector database to get documents related to the input text and add them to the prompt as context for the LLM. Although the returned documents will be semantically similar to the input text, they may not be the optimal set of documents to add to the prompt. For example, if the set of documents have duplicate content, providing the same document multiple times in the prompt will not improve the result from the LLM. In this case, you want the documents to be similar and diverse. This is where the re-ranking the results comes into play. As of release 0.1.0 LangStream includes a re-rank agent to improve the quality results that are sent to the LLM.

  • Support for Milvus as a Vector Database in LangStream

    We are happy to announce that as of release 0.0.22 LangStream now supports Milvus as a vector database. This integration broadens LangStream’s support for vector databases giving users more flexibility in their Gen AI applications.

  • Introducing LangStream: the easiest way to build streaming Gen AI applications

    Unless you’ve been living under a technology rock, you know that Generative AI is the hottest technology to come around in a while. So-called pundits are comparing it to the introduction of the iPhone, which ushered in the mobile wave, or even the World Wide Web (gasp!).