I am trying to get "letter by letter" streaming but I am not abble to make it work -- streaming answers combining Buildship & Flutterflow as per @thedigitalpro tutorials . Here below are details : thanks in advance for you help !
I implemented the Assistant Streaming Model by Buildship shown here: https://www.youtube.com/watch?v=B0g0FWqWSLg. And it does work BUT this provides a "word by word" streaming. It is not a "letter by letter" streaming like the one you can see on ChatGPT (even by making changes to the "writeInterval" I could not achieve such UX) ==> Maybe I am missing something here ?
As mentionned above, I also implemented solutions from @thedigitalpro. Those allows for "letter by letter" streaming. BUT unfortunately I was not able to make streaming work - tried both solutions here https://www.youtube.com/watch?v=HVWWEaWRgo8 and here https://www.youtube.com/watch?v=BoaxDTUwWxw ==> I checked multiple time my Flutterflow workflow ad could not find a bug there (I checked the custom action code that connects to Buildship and processes streaming, I checked Supabase and chats do come in, I cheked the arguments and action workflows into widgets and all complies with tutorials. ==> The only thing I noticed is that "threadId" does not stick in Flutterflow (shows NULL). Maybe that blocks streaming ? (Although first chat answer does not stream either and it does not require threadId) abilities.
Maybe there is something I am missing and I should deep dive into to solve my issue ?
Learn how to get streaming response from AI Assistant connected to your database and tools on FlutterFlow with BuildShip. This is a full stack template.
In this community requested video we enhance a sample chat application in FlutterFlow to support Open AI Assistant Streaming directly from a #BuildShip workflow. Streaming is common in chat style applications which slowly returns content as its being delivered from external services. This is a familiar site in tools such as #chatgpt.
If you are looking to include the use of an #OpenAI #Assistant with the added ability to save conversation history so they can be resumed at a later date then this example will demonstrate how to achieve it. Additionally this sample application also adds a little bit of complexity by supporting the use of #BuildShip Streams to replicate the effe...
Continue the conversation
Join the Discord to ask follow-up questions and connect with the community
B
BuildShip
Join founders, builders, devs using Low-code / No-code to create backend, APIs, scheduled Jobs, automation, AI workflows