Editor's Note: We're very excited to share this work by BCG. We've worked closely with the BCG over the past year to help support companies bring GenAI initiatives into production. We were intrigued to hear about their AgentKit platform, and once we got a closer look we were even more excited. We hope you all are as well!
BCG X is excited to announce today's release of AgentKit, a LangChain-based starter kit to build constrained agent applications. We frequently use LangChain inside BCG X to build GenAI applications and LangSmith to evaluate them. With AgentKit, we’re contributing back to the community by releasing something that has been useful for us in building agent-based applications for our clients.
AgentKit is a starter kit to build full-stack constrained agent applications. It is built on the latest versions of Next.js 14, FastAPI (Pydantic 2.x) and LangChain for optimal performance, security, and developer experience. Developers can use AgentKit as a foundation for your projects to:
We chose to build AgentKit on top of LangChain for a couple of reasons. LangChain enables us to both work with any LLM that is right for the application, and to natively integrate with LangSmith so that we can easily evaluate the app’s performance. Moreover, it has made it easier for us to integrate helpful features such as memory management and an extensible AgentExecutor.