Visual Product Search
A customer sends a photo; the workflow reads any printed code first and only calls a vision model when it has to.
Problem
Customers photograph a product tag and ask 'do you have this' instead of typing a code or name. Matching that by hand meant a staff member squinting at a low-res photo.
Before
Photos landed in the same inbox as text messages and waited for someone to visually match them against the catalogue.
After
Product codes are read and matched automatically. Photos without a readable code are identified by colour, pattern and type using a vision model — only when needed.
Interactive workflow
Click through the actual node graph
Try the cheap option first
Most product photos already have a code printed on the tag. Reading that code off the photo is fast and practically free, so the workflow always tries it first. Only when there's no readable code does it fall back to asking an AI to actually look at the photo and recognise the item — a slower, more expensive step. Trying the cheap option first keeps costs tied to the genuinely tricky photos, not to every photo that comes in.
How the two paths differ
If a code is found, the product is looked up directly — no AI needed at all. Only when there's no code does the workflow pull up the full product list and ask the AI to identify the item by what it looks like.
Impact
Photo enquiries get an answer as fast as text ones, without a proportional rise in the vision-API bill.