Lorai

What's up PH community! I'm Dylan, founder at Lorai. Building this has been a real journey lol. Two weeks ago I started Lorai as a tool to make Low-Rank Adaptations (LoRAs) with no code. LoRAs are basically generative AI "plugins" that allow the models to generate things it hasn't seen before (such as your brand assets) and before Lorai there really was no straightforward way to make them. People had to run github scripts, build docker images, or use google colabs to make these, so I thought I'd make it easier. I launched on the stable diffusion subreddit multiple times and got 1k+ upvotes, 100s of users, and 10 pro users paying $10/mo. But then I noticed that the people that were actually converting to the pro plan were trying to make LoRAs of their product or of their brand/creation so that they could generate new assets that would not deviate from their core brand. So I narrowed Lorai to focus on product builders rather than broad segments of people like "artists" or "creators". I also simplified the product a ton and made it possible to generate the assets within the app rather than having to use the LoRA file on your own. The reason Lorai is much faster + efficient than most fine-tuning gen AI platforms is that we use LoRAs instead of fine-tuning an entire model on your assets. This results in a file that is orders of magnitude smaller (3 GB -> 4 MB) and a training process that is orders of magnitude faster (2 hrs -> 10 minutes). Then we load this file into the model and allow you to generate assets w/o hassle!

Finally!!! This is so cool 😍

Cheers to your hard work paying off! Excited to see your product's journey.