Build AI App Prototypes Visually with No-Code (Open-source)
Вставка
- Опубліковано 9 чер 2024
- Earlier this year, Ironclad open sourced their Graphical Development Environment for AI applications Rivet. It is a great environment to prototype AI apps without any code. Especially if you’re building complex apps with multiple services, it is great to have the visual aspect and the no-code flexibility.
You can also integrate the graphs you build in Rivet directly into your Node.js application.
In this tutorial, you'll use Rivet to develop a graph that transcribes a podcast episode, and a graph that takes your question and generates an answer using LeMUR. You'll then call these two graphs to get answers to questions given an audio URL.
Download Rivet: rivet.ironcladapp.com/
Here is the written tutorial with the Node.js integration code: www.assemblyai.com/blog/podca...
What is Rivet by Ironclad?
Rivet is a visual programming environment for building AI agents with LLMs. Iterate on your prompt graphs in Rivet, then run them directly in your application. With Rivet, teams can effectively design, debug, and collaborate on complex LLM prompt graphs, and deploy them in their own environment.
How can you use Rivet?
Rivet has a graphical interface that makes it very easy to quickly prototype AI applications. Just download it, set up your plugins and start building.
How to prototype with AI APIs and services?
Rivet is a great way to prototype AI applications. Using their graphical interface and plugins for many different AI services, you can build the prototype of an AI application with a few clicks.
00:00 Intro demo
00:21 What is Rivet by Ironclad?
00:47 Setting up the environment
02:09 Project structure
02:23 Simple transcription example
03:32 Set up project for Q&A
04:11 Transcription graph
05:38 Question answering graph
07:10 Main graph for Q&A
09:06 Example run
10:15 Finalizing the output format
12:36 Node.js integration
13:09 Watch next!
▬▬▬▬▬▬▬▬▬▬▬▬ CONNECT ▬▬▬▬▬▬▬▬▬▬▬▬
🖥️ Website: www.assemblyai.com/?...
🐦 Twitter: / assemblyai
🦾 Discord: / discord
▶️ Subscribe: ua-cam.com/users/AssemblyAI?...
🔥 We're hiring! Check our open roles: www.assemblyai.com/careers
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
#MachineLearning #DeepLearning - Наука та технологія
I've seen many no-code tutorials but often found them confusing. However, your video provided clear and comprehensible instructions, motivating me to brainstorm project ideas to experiment with.
Awesome , thank you
🔥 very useful, thanks
Amazing!!! I am using here! Thanks!!!
agreed, very useful. Thank you..
Glad it was helpful!
🎯 Key Takeaways for quick navigation:
00:00 🌐 *Introduction to Rivet and Assembly AI Integration*
- Demonstrating Rivet's no-code environment for AI application prototyping,
- The ease of querying information using Assembly AI without writing code,
- Overview of Ironclad's Rivet as a tool for integrating multiple services visually.
01:09 🔑 *Setting Up Assembly AI Plugin in Rivet*
- Setting up a Q&A project with Assembly AI in Rivet,
- Adding Assembly AI plugin and obtaining an API key,
- Introduction to the project structure in Rivet, consisting of multiple graphs.
02:32 🎙️ *Creating the Transcribe Audio Graph*
- Building a simple graph to transcribe audio using Assembly AI,
- Connecting audio data to Assembly AI's transcribe node,
- Observing the transcription process and output within Rivet.
03:39 📊 *Structuring the Full Project with Multiple Graphs*
- Deleting the initial graph to create a more complex project structure,
- Establishing separate graphs for transcribing audio and answering questions,
- Each graph functions as a separate component, akin to a function in programming.
05:33 🤔 *Setting Up the Answer Question Graph*
- Creating inputs for transcript ID and user questions in the answer question graph,
- Adding a Q&A node and configuring inputs to receive user data,
- Outputting the JSON response containing the answer to the user's question.
07:11 🧩 *Combining Graphs in the Main Function*
- Integrating separate graphs into a main graph to function collectively,
- Prompting user inputs for audio URL and question within the main graph,
- Utilizing subgraph nodes to link the transcription and question answering processes.
09:36 🛠️ *Troubleshooting and Executor Switching*
- Encountering and resolving an error during the graph execution,
- Switching from the browser executor to the node.js executor for better plugin functionality,
- Successfully obtaining a JSON response with the answer to the user's question.
10:49 📑 *Parsing and Formatting the JSON Response*
- Extracting the answer from the JSON response using an object path node,
- Parsing the response to display the answer to the user's question clearly,
- Running the main graph again to demonstrate the improved answer formatting.
12:43 📚 *Integrating Rivet Graphs into Node.js Projects*
- Discussing the possibility of using Rivet graphs in Node.js projects,
- Mentioning a tutorial for further learning on integrating graphs into code,
- Emphasizing the importance of prompt engineering in AI application development.
Made with HARPA AI
Im doing something simiiar but im using 4 different AI programs in a circuit to build an application.Im doing the whole thing with English text prompting. One builds on the other and explains to me what going on line by line.Each Ai using their own angle to optimize the code..In the future coding will be done that way i belive.They'll be more programs managers then programmers
Thank you. ❤🍕🙏
Thanks for making these tutorials not intimidating 🎉
Ty for the video how is rivet compared to flowise ? Thanks
Really nice 😃.Thanks for sharing. But need to wait ~40 seconds to get answer. (Noticed runtime at right bottom corner. Time: 12:40). It would be good if it is improved
with this example, is there a way of opening a dialog box to choose an audio file, rather than typing in a url?
super
Make a video on how to make an dubbing ai like
Dubvers etc.
Plz
What was your own channels name on UA-cam?
It’s Misra Turp. :)