The new infrastructure libraries make moving a locally running generative AI stack to the cloud easier.
“The GenAI tech stack is new and emerging but has typically consisted of a LLM service and a vector data store. Running this stack on a laptop is fairly simple but getting it to production is far harder. Most of this is done manually through a CLI or a web console, which introduces manual errors and repeatability problems that affect the security and reliability of the product,” Pulumi explains.
New libraries
The company is, therefore, introducing new libraries that are the “fastest way to learn and build Infrastructure as Code.” Pulumi’s technology is designed to address the complexity of cloud infrastructure provisioning and management. Pulumi has announced collaborations with Pinecone and Langchain to simplify generative AI further.
The integration with Pinecone means that the Pulumi platform now provides functionality to manage Pinecone indexes. This serverless vector database enables developers to build and deploy high-performance AI applications. A vector database is especially useful for large language models, generative AI and semantic search. In addition, Pulumi offers a template for running Langchain LangServe in the container management service Amazon ECS.
Pulumi offers an Infrastructure as Code (IaC) platform. The platform is designed to develop modern applications quickly and effectively. However, there are some challenges as the cloud has specific developer requirements. The new infrastructure libraries address these obstacles and simplify application development.
Also read our story explaining Pulumi’s technology: Pulumi takes Infrastructure-as-Code to the next level