Describe the solution you'd like
Chainlit is great and it's also nice that already comes with integrations out of the box e.g. for Langchain. But recently I have transitioned away from Langchain and found Microsoft Semantic Kernel much better suited for productive use cases. It's probably subjective, but it looks more polished... It also has really nice features out of the box: function calling abstractions, RAG and other memory and the planner! The planner is a sophisticated component that can use the tools (functions) to come up with a formal, multi step plan on how to perform a complex task. It is really useful in case you are building an enterprise chatbot which usually requires multiple backend functions to realize the end users' request.
Describe the solution you'd like Chainlit is great and it's also nice that already comes with integrations out of the box e.g. for Langchain. But recently I have transitioned away from Langchain and found Microsoft Semantic Kernel much better suited for productive use cases. It's probably subjective, but it looks more polished... It also has really nice features out of the box: function calling abstractions, RAG and other memory and the planner! The planner is a sophisticated component that can use the tools (functions) to come up with a formal, multi step plan on how to perform a complex task. It is really useful in case you are building an enterprise chatbot which usually requires multiple backend functions to realize the end users' request.
Additional context Semantic Kernel: https://github.com/microsoft/semantic-kernel My (currently C#) demo on the features of Semantic Kernel: https://github.com/bbence84/semantic_kernel_copilot_demo