ฝัง
- เผยแพร่เมื่อ 7 ส.ค. 2024
- 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.
Cloneables here:
BuildShip Template: buildship.app/remix?template=...
FlutterFlow Marketplace App: marketplace.flutterflow.io/it...
Connecting BuildShip to your existing Firestore database: docs.buildship.com/tutorials/...
Learn how to stream AI responses into Flutterflow projects using BuildShip and Firebase, addressing a highly requested feature from the community. The video details step-by-step instructions on setting up the necessary components, including generating API keys, setting up Firebase, and configuring Firestore database for storing chat histories. Additionally, it covers deploying the Firestore rules and connecting the BuildShip workflow to Flutterflow, ultimately showcasing how to create an interactive chat-like AI assistant that enhances user engagement with real-time responses.
Chapters 📚
00:00 Introduction to Streaming AI in Flutterflow
00:48 Setting Up Your AI Assistant with BuildShip
01:36 Configuring Firebase for AI Streaming
03:55 Integrating AI Assistant into Flutterflow
04:48 Testing and Deploying Your AI Assistant
06:06 Conclusion: The Future of AI Assistant Development
Let's connect 🤗
Chat: / discord
Tweets: / buildshipapp
LinkedIn: / buildship
#aiworkflow #nocode #lowcode #backend #apibuilder #cronjob #scheduledtask #backgroundjob #openai #ai #googlecloud #aiassistant #flutterflow @ - บันเทิง
a new area opens up🥰
Awesome! Thanks a lot for this!
Glad to hear 🎉
pretty impressive ! Nice job
Thank you 💜 glad to hear you enjoyed it
A firewalled AI assistant that can't be snooped is going to be a huge hit. The rub is making it a one-button installation.
So i'm sure this is a newbie question, but I've set everything up (I think correctly) and the Firebase all looks great - but the passing of the variables from Flutterflow to BuildShip seems to not be occurring correctly. In the video I've slowed this part down and watched your steps but it looks like you already had variables in your BuildShip nodes. Am I missing something?
Given how new the low-code space is, we’re all newbies here. 🙂
Great news! FlutterFlow just announced support for streamed API responses, eliminating the need to use Firebase as an intermediary (unless you prefer to). We’ve published an updated version of the video to cover this change: th-cam.com/video/x4kCHaL_bLk/w-d-xo.html
If you have any other questions or need further assistance, feel free to reach out on Discord. We’re here to help!
Could you please make an alternate tutorial on just this topic again but configuring to Supabase? I would like to stream responses from ChatGPT-4o into FlutterFlow based on my end users data within my Supabase end users table. I cannot seem to find any information on this anywhere! Thank you for your visually engaging content!
It's great that you're enjoying our content!
We have a Supabase version of the above that you can remix here:
buildship.com/templates/streaming-assistant-supabase
Keep in mind also that FlutterFlow have just announced support for streamed API responses, eliminating the need to use Supabase as an intermediary (unless you prefer to). We’ve published an updated version of the video to cover this change: th-cam.com/video/x4kCHaL_bLk/w-d-xo.html
Great video! How would I go about giving the chat user the option to include an image upload into the chat to serve as context as well?
Super helpful and many thanks for putting together!!! I noticed my custom instructions in OpenAI aren't flowing through. Should I be setting up my custom instructions in buildship using this approach? Thanks again!
Correct, the custom intrusions on OpenAI will be overwritten by OpenAI assisstant node instructions. We recommend simply setting up the assistant on the OpenAI console and then setting up everything else on BuildShip.
great job, thank you!
but im personally experiencing the issue, assistant answer is not appearing in firebase, its always empty. How to resolve 😢
Greetings to the buildship team, good job. One question, are you preparing some type of template or nodes for when gpt4o comes out with camera vision and audio to implement it with Flutterflow? It would be something very differentiating on your part and I think you would eliminate all your competition apart from helping us. all the best
Yes 100%, as you might have noticed we launched updated versions of the node even the same day of the launch. Also you have full flexibility to make changes to the node as well - this gives you the power to move fast :)
Thanks for an excellent job
Would it be possible to do the same with Supabase?
If I ask the Buildship AI to redo the node with Supabase, do you think it will work?
Your wish........
buildship.com/templates/streaming-assistant-supabase
Keep in mind also that FlutterFlow have just announced support for streamed API responses, eliminating the need to use Supabase as an intermediary (unless you prefer to). We’ve published an updated version of the video to cover this change: th-cam.com/video/x4kCHaL_bLk/w-d-xo.html
Great tutorial! Could you make one on a chat with streaming and saving the chat history?
Hi 👋this one already maintains history with threadID and also because it writes to Firestore you have history there as well.
@@BuildShipApp omg, nice one 👍 memory is the most important requirement for an assistant
this is very interesting. can you create a tutorial using supabase instead, or do you think it'd be better to use firebase for this function? loving buildship btw :)
Your wish........
buildship.com/templates/streaming-assistant-supabase
Keep in mind also that FlutterFlow have just announced support for streamed API responses, eliminating the need to use Supabase as an intermediary (unless you prefer to). We’ve published an updated version of the video to cover this change: th-cam.com/video/x4kCHaL_bLk/w-d-xo.html
Hey, this looks really great, but I have a question. Are there any plans to make this new node also work with supabase?
Your wish........
buildship.com/templates/streaming-assistant-supabase
Keep in mind also that FlutterFlow have just announced support for streamed API responses, eliminating the need to use Supabase as an intermediary (unless you prefer to). We’ve published an updated version of the video to cover this change: th-cam.com/video/x4kCHaL_bLk/w-d-xo.html
Awesome! Can we use 4.o as the model in our assistant with these nodes?
Sure can. 🙂
Yes this is already available in this assistant now. You can get a fresh copy from the library in case your previously added to see the GPT-4o support.
Thanks for this! My Assistant is a simple chat back and forth. No website scraping. When I enter a prompt and the Assistant responds, the chat is captured in Firestore collection just fine. But I don't understand how to get the threadId back to FF. I tried adding a node after the Open AI Assistant streaming node, but apparently I didn't explain to the AI what it should do very well because the result crashed the workflow. How do I get the threadId back to Flutterflow?
Never mind. Earlier I added a return node and when I clicked on javascript, I kept getting an error. I just tried it again by adding threadId and it worked as expected. Don't understand why I was getting the error though, which was something like function failed or problem with editor.
Glad to hear it’s resolved. Happy to support via in app support form in case of any questions in the future :)
Most of these middle-man services are optional, especially considering security (apikey) and handling scaling.
One can absolutely achieve this with just cloud functions. Efficient, fast, and less costly if your functions are well written.
And you don't have to handle keys on the client side platforms.
This is backend api and not client side. Your key is securely stored on Google cloud project secret manager and not deployed with the app on frontend on flutterflow - this is also scalable on GCP better than cloud functions. This AI assistant builder makes the process of building and deploying easy and nocode friendly with ready to use integrations to speed up development. Ofcourse anything can be achieved directly via code if you are familiar with code and you can build your own backend from scratch.
BuildShip is aimed for those who want to build fast and efficiently at the same time retain the flexibility to change code if needed as BuildShip makes the code accessible and deploy to your own cloud if you would like as you grow and scale if you choose. Fully flexible and extensible :)
I do not need thread, only text generation, how do i modify it for this? My application: I send a key word and it generates text output
Swap out the Open AI Assistant node in BuildShip for the Open AI Text Generation node and that should give you what you're looking for. If not, reach out to our support team or join our Discord community where we can help you in more detail.
I'm just wondering if this would increase the read and write cost of the firebase.
It’s usage based just like how you will read and write from firebase from any app.
My API call is working, because I can see data being written to Firestore, but nothing is updating in the FlutterFlow UI when I test it. It keeps the same boilerplate text "Streaming AI Assistant Chat App Template using FlutterFlow and BuildShip...". What am I missing?
Hi there, have you connected the API on flutterflow app with your shipped API? If any issues raise an in app support and we can help you out.
@@BuildShipApp Thanks for the response! The API was indeed connected on flutterflow. I reloaded the test session several times and eventually it started working 🤷♂
How do I edit/get rid of that boilerplate text on initial load?
@@JamesPurcellMedia In the widget tree, edit the component titled "BlankListComponent"
Hello, I follow all the steps to have a functional chatbot but I'm experiencing a prompt that sometimes incomplete, what seems to be the problem?
@@johnpaulobattung7438 hi is your prompt incomplete or the response? One easy way is to first test the workflow or individual nodes to ensure every step and the entire workflow works and then once that’s good - you can test via flutterflow to ensure integration works. This helps isolate and identify if any issue. Also you Could you raise an in app support form and we can take a look?
@@BuildShipApp hello thankyou for replying to my concern, the response from AI is incomplete. I'm not sure if this from buildship or flutterflow even though I get always success creating thread ID.
I can raise an app support, in Buildship right?
its a chat bot. Another literal clone of an already existing conversation model serving absolutely no new purpose for no new problem
Hey you might have missed the point :) these are not just chatbot but are connected to an AI assistant workflow that you can use with any model from OpenAI, Groq, Azure, Claude and any tool like Notion, Stripe, Database and have conversation with memory, thread and streaming. So it’s much more than a regular chatbot, you can let your chatbot take actions, give meaningful answers connected to your database. Check out our AI Assistant playlist to explore it. Thank you.
are these features free or is there any cost for someone who is building an app using flutterflow and wants to implement this AI assistant in it?
BuildShip has a free tier on which you can build and try it our fully free. If you want to scale and not have rate limit etc then you can upgrade to paid plan as you need