A mobile operator has decided that the thing standing between customers and its branded hoodies is the effort of putting one on. T2 is testing AI-powered virtual fitting rooms in its stores, and the early numbers are modest but not nothing: about a thousand successful fittings since August, followed by 150 purchases.
How the fitting works
The process is short, and nobody has to create an account. The report says no prior registration is required.
- Pick an item. The visitor chooses a garment with T2 branding on a virtual showcase.
- Take a photo. The built-in camera captures the customer.
- Enter three numbers. Gender, height and weight.
- Wait about 30 seconds. The algorithm works out the right size and shows how the garment would look on that person.
There is no mirror and no changing cubicle, and nobody asks whether you would like to try a different size. The size recommendation is part of the output, so the machine does the job of the helpful shop assistant, minus the small talk.
The numbers, and what they can and cannot say
The pilot started in August and will run until the end of 2026. The equipment is installed in more than 50 T2 stores across Russia.
The headline figures are about 1,000 successful fittings and 150 purchases afterwards. A back-of-the-envelope reading puts that at roughly 15 percent, which is a respectable figure for a kiosk that most people walked past on their way to fix a SIM card.
Treat that ratio with caution. The report does not say whether each purchase maps to a single fitting, how many people tried several items, or what the sales of these items looked like before the kiosks arrived. Without a baseline, 150 sales is a number, not a verdict.
What T2 says happens next
When the pilot ends, T2 plans to assess whether the project could be extended to more stores in the regions where it operates. In other words, the decision is deferred until the data is in, which is the sensible way to treat a kiosk.
The company's pitch is simple: customers can judge how clothing looks without physically putting it on.
What the report leaves out
The source is short, so several obvious questions go unanswered:
- Technology. Nothing is said about which model or vendor powers the try-on.
- Photo handling. The report does not say whether the captured images are stored or deleted.
- Accuracy. There is no data on how often the recommended size turned out to be right.
- Pricing and range. The report mentions clothing with T2 symbols but gives no detail on the catalogue.
Until those answers appear, the pilot is best read as a small retail experiment, not as a verdict on virtual fitting. It is a telecom operator selling merchandise, a camera and an algorithm standing in for a mirror, and a company that wants to count the receipts before deciding whether to buy more kiosks. That is a refreshingly modest way to adopt AI.