The idea

I've spent the last year exploring AI infrastructure services like Replicate and OpenAI, using their technology to create my own apps.

One key thing I found out is how combining these powerful models can tackle even more complex tasks.

However, using and tweaking these models usually needs tech skills, like coding, which not everyone has.

That's where Nodeline comes in. It's a No-Code platform designed to link together "nodes", making it easier for anyone to use. Each node has its own inputs and outputs, and can be set up with an AI model for specific tasks. You can also find nodes for different purposes like text, YouTube videos, or file uploads.

Nodes

Each node is like a building block, with inputs and outputs that let you chain them together into sequences.

Here are a few node examples:

• YouTube Video

  Downloads a YouTube video in audio or video format and uses that content in other nodes.

• Transcription AI

  Converts audio to text using OpenAI's Whisper.

• GPT-3.5

  You can use GPT-3.5 for processes that require the "reasoning" of a large language model for more complex tasks.

Pipelines

Imagine creating a pipeline that can summarize YouTube videos in just a few clicks, using the nodes mentioned above:

First, we download the audio of the video we want to summarize. Then, we connect that node to the transcription node to convert the audio to text. The transcription node outputs the generated text, which we can use with GPT-3.5 to create the summary.

Reusable & Modular

This pipeline is reusable, and you can repeat the task as many times as you want with little effort. Also, being modular, you can replace nodes and create new pipelines. An example could be, going back to the previous example of the YouTube video summarizer, we could replace the YouTube node with the file upload node and use the same pipeline with our own files.

Thanks!

If you like the idea of ​​Nodeline, follow me on Twitter (@markdoppler_) for more updates.