I’ve been experimenting with the Assistant API for a few weeks now, and am ready to start putting some project ideas into production. First I want to make some chatbots.
I’ve messed with VoiceFlow’s implementation and even used their API blocks to interact my own Python+Flask instance of an Assistant, but I don’t really see the need to pay for their service if it’s basically just a frontend component.
What do you use for building/hosting your chat window?
Do you prefer something for the backend different from Python+Flask?
What do you like to use for micro service APIs, when knowledge docs are not enough? Again I’m thinking Python+Flask spun up on something easy like Replit.
The beauty is that the API can be used with anything. Some people use ruby, some people use R, some guy made a fortran lib
I’m personally a fan of node, but you sometimes do need to use python for some stuff. If you’re comfortable with it, it’s a very versatile option.
Regarding front ends, I’m starting to wonder if front end libraries are starting to lose their purpose - it’s so easy to whip something up from scratch with gpt-4/copilot with a little bit of practice.
And docker to tie it all together. Wrap everything into an image and you can run it anywhere.
But imo the choice of stack is person/team/project dependent.