I built a whole wardrobe system in ChatGPT + Notion and it’s gotten way more useful than I expected
retained text
I wanted ChatGPT to actually know what I own instead of giving me generic outfit ideas, so I ended up building a wardrobe system around it in Notion. The basic goal is to get the most outfits out of the smallest useful wardrobe and remove as much daily decision-making as possible. Notion is the source of truth, so every shirt, pair of pants, shoes, watch, jacket, etc. has its own record with things like color, fit, formality, season, current status and whether it’s actually available to wear. The daily outfit part is probably what I use the most. It can look at the weather, what I’m doing that day, what’s clean or dirty, what I’ve worn recently, my saved outfits, and my watch/accessory rotation before recommending something. I also added outfit “archetypes” so it doesn’t have to search every possible combination from scratch. It can first decide that today calls for something like a warm casual outfit or a polished transitional outfit, then fill that structure using clothes I actually own. Successful combinations get saved to an outfit library instead of filling the database with thousands of theoretical combinations. The part that made the system much better over time was treating actual wear as evidence. If I wear something and love it, hate it, find it uncomfortable, realize the colors look dead together, or get feedback from someone else, that can be logged and eventually turned into a real styling rule if the pattern keeps showing up. There’s also a failure log for the system itself. If ChatGPT makes a bad recommendation because it ignored the weather, misunderstood a garment, used something that was dirty, or repeated some other preventable mistake, the failure gets diagnosed and a prevention rule can be added instead of just moving on and making the same mistake later. Shopping runs through the same system. Before buying something, I can have it check whether I actually have a wardrobe gap, what outfits the new item would unlock, whether I already own something that does basically the same job, and whether the purchase adds enough utility to justify taking up space. I also have it consider things like free shipping when it makes sense instead of evaluating every item in isolation. Returns, trial items, tailoring, repairs and stuff I’m getting rid of all have their own statuses, so the inventory stays operational instead of becoming a dead closet spreadsheet. Travel planning ended up being another surprisingly useful feature because the system can plan several days at once while accounting for weather, activities, rewearing, laundry state and packing space. I also have an AI operating manual inside Notion so a completely fresh chat can cold start, read the rules and databases, and continue using the system without me having to explain the whole thing again. The image side is still the messiest part. Once an outfit is decided I like being able to visualize it, but image generation has a tendency to substitute or invent clothes, so I’ve mostly separated “decide the outfit” from “render the outfit.” The actual wardrobe database gets final authority. At this point it feels less like asking ChatGPT what to wear and more like having a little wardrobe operating system that happens to use ChatGPT as the reasoning layer. The best part is probably that it gets more useful from real wear instead of just producing endless new recommendations.
document fetched 2026-10-20 00:21 · canonical https://reddit.com/r/ChatGPT/comments/1x9kyow/i_built_a_whole_wardrobe_system_in_chatgpt_notion