How I Used AI to Simplify My Embroidery Projects
Heads up: Whether you’re experienced in AI or embroidery — or completely new to both — this post has something for you!
The struggle
In the embroidery world, there are many thread brands and a looooot of options for the same color. Imagine you need the perfect green for your project. There are countless options, varying in shades, brightness, and more. Even within a single brand, you can easily have dozens of different options.
I always struggle when deciding which color to pick for a new project. Every time I’m in the atelier and glance at the wall full of threads, I feel overwhelmed by the endless choices.
BordAI
Being passionate about manual arts and technology, I came up with the idea of creating my own personal assistant for embroidery.
I called this assistant BordAI.
I developed two tools to support my embroidery projects:
- Suggest threads from the image: given an image, it identifies the most important colors and suggests the closest thread matches from the Anchor or DMC palettes.
- Thread color conversion: based on input thread codes from either Anchor or DMC, it quickly converts them to the matching colors in the other brand.
To make the experience even better, I built a data app using Streamlit integrated with the Gemini API. This way, I can chat interactively with bordAI about my ideas, and it automatically activates the necessary tools whenever it identifies they’re needed during our conversation.
Let me show you how it works:
The user sends a message, and it classifies whether it intends to use:
- Image-based thread suggestion
- Thread color conversion
- Chat
Example 1:
In this example, I took a picture of an embroidery I had already made, so I already knew the threads used, allowing me to check if the thread suggestions provided were acceptable.
The app asks if you’d like to remove the image background — in case the picture contains extra elements that aren’t relevant to your embroidery. There’s also an option to adjust brightness, useful when the image isn’t accurately representing the ideal colors.
In this case, I applied both parameters.
Sometimes, we don’t have the exact recommended thread, and image brightness can also affect the results. For this reason, the output includes the identified color and the top three most similar thread colors, ranked from most similar to least similar. This increases the chances of you having at least one of the recommended threads.
It got most of the colors! Out of 11:
- 5 were exactly the right code
- 2 were very close — just one digit off
- 4 weren’t identified, but honestly, that’s already a pretty solid start!
Example 2:
Here, I took a photo of a beautiful sunset to use as a reference for a new embroidery.
It gave me a beautiful palette for inspiration!
Example 3:
In this example, I’m demonstrating the Thread Color Conversion tool.
BordAI also has memory, so it remembers everything discussed during that session. This way, you can use the conversion tool and afterward ask related questions without losing context.
Try it yourself!
It’s free and open to everyone — check it out here: https://bordai.streamlit.app
Stay tuned!
This will be a series of posts where I’ll share details about how I developed this app and its upcoming improvements. In the next posts, I’ll explain how I integrated the Gemini API, how it classifies the messages and accesses the tools I built during our conversations.
I still have many ideas from my embroidery experience, and I’m excited about combining my hobby with my professional skills.
I hope this inspires others to explore AI solutions to solve their struggles too!
I’d love to hear your thoughts! Let me know in the comments what you think, or if you have any suggestions.
