ThreadIt

login
dashboard
editor
ai generated photo
shopify deployment
shopify storefront
Image of final storefront

Inspiration

To get inspiration for what we wanted to build, we started by calling up a lot of our friends, asking them if there was anything in their life they wished they could hand over to AI. A couple of phone calls later, we talked to one of our friends who is trying to get into fashion and start his own clothing brand, but he isn't the best graphic designer; he says it sometimes takes hours to make even simple designs, and putting them out there is sometimes a hassle. We thought, “let’s automate that.”

What it does

ThreadIt takes a rough sketch and turns it into a polished clothing design that appears on Shopify within minutes. You open a canvas, draw on top of a garment template, click a single button, and the app captures your art, feeds it through an AI pipeline for visual upgrades and product copy, then posts a zero-inventory wait-list item to your store. From idea to live listing, everything happens in one uninterrupted flow.

How we built it

We started with a React front end because it let us move fast and plug in TLDraw for a smooth drawing experience. Zustand keeps the canvas and preview state in sync, while Vite handles hot-reloads that never break the session. On the server, we wrote an Express app that accepts multipart uploads, pipes them through Sharp for a quick resize, then calls Stability AI for image generation and enhancement. Claude writes titles, descriptions, and tags. When the final file is ready the server hits the Shopify API, creates a product, uploads the image, and installs a branded theme so everything looks consistent.

Challenges we ran into

Image generation at first was completely off. Shirts were coming out as animals, patterns were floating in mid-air, and nothing looked remotely wearable. We tried everything: Gemini was inconsistent, DALL·E gave us too much abstraction, and Hugging Face models lacked the control we needed. Eventually, we landed on Stable Diffusion paired with controlNet for more predictable output. It still took a lot of trial and error, prompt tweaking, strength adjustments, and parameter tuning, but we finally got it to produce clean, product-style images.

Accomplishments that we're proud of

Within a day, we were able to go from simple outlines to a live Shopify listing with professional-grade imagery and copy in under sixty seconds. The entire process is automated, letting users avoid Photoshop and Shopify configuration altogether.

What we learned

Stable Diffusion rewards careful parameter tuning far more than brute-forcing higher resolution. Prompt engineering for apparel is its own art, terms like “studio lighting,” “cotton texture,” and “shallow depth of field” consistently lift quality. We also learned how to juggle multiple asynchronous AI calls without blocking the user experience and how to keep large images within Shopify’s limits without killing detail.

What's next for ThreadIt

Next we plan to add 3D garment mockups so designers can spin their creations on virtual models and see exact drape and fit before ordering samples. We are also building a matchmaking feature that connects creators to manufacturers who fit their fabric, quantity, and sustainability needs, closing the loop from concept to production