I'm planning to develop an AI app tailored for a specific use case. Basically, it's going to be a wrapper around GPT, utilizing a language model with retrieval-augmented generation (RAG) and distinct memory for each user, plus possibly some tool calling. I'm having trouble finding a unified backend solution that fits all these needs. What tech stacks do you guys recommend?
6 Answers
You should definitely check out AWS Amplify. It's super easy to set up and budget-friendly. Plus, you can choose the front end that suits you, and as you scale, integrating other AWS services is a breeze.
I'm officially using Go for my backend, but honestly, I'd suggest Node.js. It comes with a lot of libraries that can make your life easier as you build.
For starters, I recommend using the OpenAI API with pgvector. It keeps things straightforward and simple, just right for getting things off the ground.
Thanks for all the great suggestions, everyone!
That sounds like a fascinating project! You might want to explore combining LangChain or LlamaIndex for RAG and memory management. A tech stack involving FastAPI or Node.js could effectively manage user sessions and tool interactions. Plus, hosting platforms like AWS or Vercel can handle your deployment easily. There's no perfect solution yet, so a modular approach to your stack might be the way to go. What tools are you thinking of using?
Take your time selecting! It's essential to find a combination that aligns well with your app's needs.
Depending on how unique your application is, you might want to consider building your own solution. It could give you more flexibility to cater to your specific requirements.
I'm considering a few options, but I'm still not sure what fits best for my project.