Custom Knowledge, Train Your Own Image Model, More!

Another big update for our faithful users! Here’s a quick list of what you just got access to:

• NEW: Document Editor
• NEW: Personas and Workspaces with Custom Knowledge upload
• NEW: Implement Flux LoRA
• NEW: AI Academy Banner on the Dashboard

Some of these are very advanced features, and more detailed tutorials will be coming soon. In the meantime, let’s break each one down briefly.

New Document Editor

Our new document editor is beautiful. With more intuitive rich text formatting features, this is the minimal-yet-highly-functional text editor you’ve been waiting for.

Not only do you get formatting options in context, only when you need them, we’ve also added an AI generation feature that will allow you to make AI edits and additions right inside the document.

iframe

You can even upload images into your documents, making it perfect for writing blog posts or ebooks.

Add “Knowledge” to Personas & Workspaces

Sometimes you want your AI to have access to consistent sources of truth, and not have to jog its memory every time you start a new conversation. Or maybe you want to have specific personas that are trained on specific data, details, or standard operating procedures.

This is where “Knowledge” comes in. You now have the ability to upload files directly to your Workspaces and Personas, giving them permanent access to the contents of those files.

iframe

The best part? This knowledge doesn’t consume “memory” in the same way that adding a file to a conversation does. This knowledge is stored in a private database that only your Workspace or Persona can access.

The possibilities here are very exciting! I can’t wait to see all the ways you use this new super power.

And just because I love a good teaser: this is phase 1 of a big upgrade roadmap for Personas in particular. Next steps will be a Public Persona Marketplace, Custom Functions/Abilities, and Agentic workflows.

Stay tuned. 😉

Train Your Own Image Model

Have you ever wanted to be able to train an AI image generator on your face? Or maybe you want to train it on a particular style you love? Or maybe you want to train it on your product… this is now possible with Magai’s Flux LoRA training.

iframe

LoRA, or Low-Rank Adaptation, is a technique used in AI image generation to fine-tune pre-trained models with a small amount of new data. By focusing on specific features or styles, LoRA allows for efficient personalization without requiring retraining of the entire model, thus enabling the creation of customized images while preserving style or subject details.

In simplest terms: LoRA is the term for adding custom training to an image model.

AI Academy Banner

In case you haven’t heard, Magai has teamed up with The AI Academy to offer our users free access to their growing online learning hub. I’ve been joining the founder, Steve Dotto, weekly for live training sessions, teaching in-depth about the latest AI tips, tactics, tools, and advancements.

So now, just in case you haven’t registered, you have a friendly reminder right inside your dashboard.

And that will do it for this update! Back to work, friends. ☕️