LongCut logo

Plexe AI Demo

By Plexe AI

Summary

Topics Covered

  • Plain English Replaces ML Engineering
  • Full Code + API + Diagnostics Together
  • Agent Team Replaces Your ML Team

Full Transcript

Plexi is an AI data scientist that builds machine learning models from plain English. It connects to your data

plain English. It connects to your data and can also help you uncover deep insights about it and build dashboards.

You can connect your data to the platform using one of our built-in integrations or via file uploads.

After connecting your data, you'll get a quick preview and you can start using this to build machine learning models.

Here I'm using the Plexi agent to analyze some of my fraud data and as you can see it's created some visualizations so I can get a better sense of the data.

Now I'm ready to build a machine learning model for fraud detection. So I

would ask Plexi to build me a fraud detector.

Plexi acts as a thought partner and if it doesn't understand something about your data, it'll ask you questions and then once you've clarified everything, it'll kick off the model build process where it'll experiment with different

types of architectures, find the best performing model and then deploy it to an API endpoint.

Once your model has been created, you get an API endpoint that's deployed and available for you to start using.

You get a thorough evaluation report so that you can have confidence in the model that's been built by Plexi.

As you can see, we go above and beyond just performance metrics. There's a

baseline comparison and a series of diagnostics that run automatically.

To give you a quick preview of what happens underneath the hood, Plexi uses a team of machine learning engineering agents to analyze your data and work like a machine learning team

would to build your model.

After it's built the the best performing model, you also get all the code available for it. So you can use this in your own internal systems. Along with these u you can also use

plexi for building dashboards via the chat interface.

This is a quick example of a dashboard that I built for the fraud detection model.

Thank you for watching.

Loading...

Loading video analysis...