We started piloting Sinan AI on Biscotto's Shopify store 48 hours ago. Biscotto sells homemade cookies and tiramisu, mostly to repeat local customers and some gift orders. It's a small store, which is part of why we picked it: small stores are where most of our actual customers live, and we wanted to see what Sinan AI does on a store that looks like theirs, not on a store we built to make Sinan AI look good.
Two days in, we don't have results worth reporting. We have a setup, and we want to write down what we're tracking before we start seeing patterns, so we're not tempted later to pick the metrics that make the story look better after the fact.
What we're watching
- Traffic: where visitors are coming from, and whether that changes once Sinan AI starts responding to questions on the site.
- Response time on inbound questions, compared to how Biscotto's owner was handling them before, which was mostly by hand, whenever she had a free minute between orders.
- Conversion from a question asked to a completed order, over the same product mix Biscotto already sells. We're not adding new products or promotions during the pilot. We want to isolate what Sinan AI changes, not what a sale changes.
- Cases where Sinan AI gets it wrong or has to hand something off to Biscotto directly. We plan to write about those too, not just the parts that go well.
Why 48 hours isn't enough
A cookie and tiramisu business has real seasonality and real day-to-day noise. Two days could be a fluke in either direction. We're not going to draw conclusions from a weekend of data, and we'd be skeptical of anyone who did. What we can say is that the pilot is running, Sinan AI is live on the store, and we're logging everything as it happens rather than reconstructing it later from memory.
Why we're doing this in public
Most of what gets published about AI tools for small businesses is either a vendor's case study, written after the fact with the numbers that flatter the vendor, or nothing at all, because the business is too busy running orders to write anything down. We wanted a middle option: a real pilot, tracked as it happens, with a business that has no reason to make Sinan AI look good other than wanting it to actually work for them.
Biscotto agreed to this because the owner is curious whether Sinan AI saves her time answering the same questions about ingredients, pickup times, and order sizes over and over. That's a narrow, practical question, and it's the one we're trying to answer honestly, not a broader claim about what Sinan AI can do for every small business.
We'll post again once we have a week or two of data, including the parts that don't go well. If Sinan AI slows Biscotto down, misreads a customer, or doesn't move conversion at all, we'll say that too.