9 April 2023
AI arrangement pricing: the 30-second quote
A florist used to spend ten minutes on every walk-in arrangement quote. Snap the bouquet, type the occasion, and our vision model returns three price tiers and the stem breakdown in under a minute. Here's how the model is built.
The thing it replaces
The walk-in customer points at three flowers in the cooler and says "something like this, but bigger, for AED 250?" The florist now has a choice: spend ten minutes pricing it from scratch, or guess high and lose the sale.
We're not the first people to think a vision model could help here. We are, we think, the first to ship one tuned on actual GCC stem pricing.
How the pricing engine works
Three layers:
- Vision pass identifies stems, count, and arrangement type. Trained on 6,400 labelled bouquets across UAE shops. Hydrangea, ranunculus, garden roses, tulips, anthuriums — the GCC top-50.
- Cost lookup pulls the shop's own current per-stem cost from inventory. Yours is yours; we never blend across tenants.
- Margin engine applies the shop's configured tier rates and produces three price tiers — Lovely / Beautiful / Stunning — with the stem deltas spelled out so the customer can pick.
The whole call returns in under a minute. Each query costs the shop one AI credit (Professional tier = 1,000 credits a month).
Why this is hard
Stem prices in Dubai vary 40% week to week — Holland flight delays, Eid demand spikes, currency. The model has to use the shop's current cost, not last week's. Pricing across photos taken with five different phone cameras under three different light conditions has its own challenges.
Try it on a real cooler photo: [/pricing-ai](/features).