AI workspace for crochet and knitting · Working Lovable product · 2026
YarnSmith
I know crochet well enough to know when a confident answer is dangerous. I designed YarnSmith to help makers analyse photos, check patterns, calculate gauge and match yarn without pretending estimates are facts.
The working product lives in Lovable. Pattern Builder is still coming; the other tools, authentication, profiles and saved workspace work now.

The challenge
Crochet and knitting decisions are full of incomplete information. Photos do not reveal exact gauge, patterns can be unclear, and a technically equivalent yarn may still be completely wrong for the person or project.
My role
I conceived and designed the entire product, including the tools, flows, profiles, workspace, editable results, AI boundaries and visual direction. I used AI as a design partner, then Lovable implemented the product.
What changed
A working public product now connects photo analysis, pattern checks, gauge calculation and yarn matching with personal preferences, saved work and results the maker can correct.
01I designed it from the craft outward
I crochet a lot. I knit too. I know that a yarn label, a pattern and a project photo never tell the whole story. Fit, fibre, drape, softness, gauge, local availability and the person wearing the thing all matter.
That domain knowledge shaped the product. YarnSmith is not a chatbot waiting for a clever prompt. It offers focused tools for the jobs makers actually need to finish.

02Not a pile of disconnected AI tricks
Authentication, profiles and the workspace turn the tools into one product. A maker can save analyses, keep different profiles and reuse preferences instead of explaining the same constraints every time.
Profiles include region, measurements, currency, size, fit, preferred brands and shops, plus practical constraints such as vegan, wool-sensitive, non-scratch, machine washable and baby safe.


03Useful estimates, not fake certainty
A photo can suggest stitch family, yarn weight, palette, fit and approximate yardage. It cannot produce divine truth. YarnSmith says the result is an estimate, shows realistic ranges and lets the maker change the assumptions.
That correction loop matters more than making the AI sound impressive. If the maker changes size, fit, stitch density or metres per skein, the result recalculates around information they actually know.


04A technically correct yarn can still be wrong
Yarn matching considers more than weight. Fibre, softness, warmth, drape, budget, region, preferred brands, preferred shops and project needs all affect whether a recommendation is useful.
The result explains why each yarn fits, shows where it compromises and reminds the maker to check availability and swatch before buying.
Open YarnSmith
05What works now
Photo Analyzer, Pattern Doctor, Gauge Calculator and Yarn Matcher work now. So do authentication, profiles, preferences and the saved workspace.
Pattern Builder is still coming. YarnSmith is a working Lovable product, not a claim that every estimate has been scientifically validated or that the AI knows more than an experienced maker.