October 2023 Summaries

Today, we are excited to announce our partnership as a launch partner for LangChain Templates, which provide easily deployable reference architectures for building generative AI applications. With DataStax Astra DB and LangChain Templates, developers can now quickly build and deploy applications that use large language models (LLMs) as RESTful web services. LangChain simplifies the process of working with LLMs by abstracting complex aspects and organizing individual components into chains for specific use cases. The combination of LangChain Templates and Astra DB's integration with LangChain can significantly improve the development process for LLM-powered applications.

Generative AI is revolutionizing the way businesses interact with their customers and create content. Tools like Audiostack and DataStax enable enterprises to leverage predictive and generative AI for real-time response, creating targeted high-quality ad campaigns in minutes compared to months of manual work. These platforms provide end-to-end data management solutions that handle data from various sources, vectorize it, store it, and use the data to make decisions and generate content. By combining cutting-edge generative text-to-speech with professional quality sound design, mixing, and mastering functionality, AudioStack's AI audio processing engine can automate and scale the entire audio production process, creating content faster than real-time and at a low cost.

A new integration between DataStax Astra DB and LlamaIndex has been introduced, simplifying the process of building generative AI applications. LlamaIndex is a popular data framework for connecting custom data sources to large language models (LLMs). The Astra DB LlamaIndex connector enables real-time ingestion of vector embedding data from an Astra vector database directly into an app, allowing users to create powerful, petabyte-scale, data-augmented chatbots and agents. Sample code demonstrates how to read a PDF, generate embeddings, store content in Astra DB, and use the ingested document to answer questions according to the Retrieval Augmented Generation (RAG) pattern. This integration is one of many being developed to make it easier to build Gen AI apps with Astra DB.

DataStax has achieved the globally recognized ISO 27001 certification, demonstrating their commitment to protecting customer data. The certification also covers additional controls defined in ISO/IEC 27001:2013, ISO/IEC 27017:2015, and ISO/IEC 27018:2019 standards. Coalfire Certification, an accredited certification body endorsed by ANAB and UKAS, validated the achievement. This milestone adds to DataStax' previous attestations with SOC 2 Type II, PCI DSS certifications, and a HIPAA opinion letter.

Vector search is an essential component in generative AI tools due to its ability to incorporate real-time information while avoiding hallucinations. However, selecting the right vector search product or project can be challenging given the numerous options available. Key challenges include handling high dimensional vectors, scale-out replication and partitioning, garbage collection, concurrency, effective use of disk, and composability. DataStax tackled these issues in its implementation of vector search for DataStax Astra DB and Apache Cassandra by leveraging SAI (Storage-Attached Indexing) and developing JVector, an open-source embedded vector search engine. These solutions allow developers to seamlessly integrate classic CRUD database features with vector search capabilities, improving productivity and accelerating time-to-market for generative AI applications.

The text discusses the integration of LangChain, a framework for developing generative AI applications, with Apache Cassandra® and DataStax Astra DB to simplify building GenAI applications. It highlights the need for speed and scale in GenAI apps and agents, which is addressed by JVector's implementation that supports high-growth use cases. The text also mentions the contributions of a connector to Cassandra and the introduction of CassIO as a Python library to abstract the process of accessing Cassandra. Finally, it invites users to join a live webinar on October 26 to discuss building production RAG apps using LangChain, LLMs, and vector search.