Inside Digikala’s AI Push: 1.17 Million Assistant Users and Lower Content Costs

Digikala says its AI shopping assistant reached 1.17 million unique users in the Iranian year ending March 20, 2026, while automated content production saved the online marketplace 5.213 billion tomans.

That saving was worth approximately $22,550 at the Tehran free-market rate of 231,200 tomans to the dollar on September 23, 2026.

The figures appear in Digikala Group’s annual report for Iranian year 1404, which ran from March 21, 2025, through March 20, 2026. They are Digikala’s own operating metrics and have not been independently audited.

AI moves into the shopping journey

Digikala says AI-generated review summaries were available for 424,000 products by the end of the reporting period. One million users read the summaries each month, producing 5.2 million monthly clicks, according to the report.

The company reported an 88% satisfaction rate for the summaries and said products carrying them recorded a 4% increase in the rate at which users added items to their carts. The report does not disclose the measurement period or methodology behind that uplift.

Visual search usage grew twentyfold from the previous year, the company said. Its shopping assistant, meanwhile, handled 2.52 million conversations and 7.1 million messages from 1.17 million unique users during the year.

Digiato covered the assistant’s launch in the Digikala app in November 2025. Known as DIA, it was introduced as a way to find and compare products through conversation rather than conventional keyword search.

Digikala also attributed higher revenue to recommendation modules. It said “frequently bought together” suggestions drew clicks from 140,000 users a day and doubled average daily revenue for that section. A “similar products” module reached 400,000 daily users and produced a 200% increase in average daily revenue, according to the report. It does not specify the baseline period for either comparison.

Accessibility tools reach more shoppers

The report says Digikala’s assistant can help blind users complete a purchase without viewing the screen. In one month, 515 blind users and 205,000 users with low vision used Digikala, although the report does not state how many of them completed an order through the assistant.

Another AI feature, “Hear the Product,” converts written product descriptions into audio. It was available for 2,108 products and used 250,000 times a month by 115,000 users, Digikala said. The company reported a 1.2% increase in purchase conversion for products carrying the audio introduction but did not disclose the methodology behind the comparison.

Content automation produces measurable savings

Behind the customer-facing tools, Digikala used AI in four content-production workflows:

• 22,838 AI-assisted model-photography frames saved 2.675 billion tomans.
• Translation and subtitling for 35,775 videos saved 2 billion tomans.
• AI-generated titles for 25,160 products saved 331 million tomans.
• Production of 110 complete videos saved 207 million tomans.

Together, those figures equal 5.213 billion tomans, or roughly $22,550 using the September 23 free-market rate. The dollar equivalents of the individual savings are approximately $11,570, $8,650, $1,430 and $895, respectively.

The content figures broaden the company’s AI strategy beyond the customer interface. Digikala has also announced AI-based pricing and demand-forecasting tools for marketplace sellers, although the latest annual report does not quantify their financial effect.

A wider super-app strategy

Digikala places the AI rollout within a broader effort to combine retail, rapid delivery, financial and lifestyle services in one app. It says daily super-app users rose 87% during the year, while 10% of users were using more than one service by year-end.

The report also records a fivefold rise in monthly sessions for what it calls the “45-minute service,” but does not define the service or the comparison baseline in the relevant section.

The annual report itself also includes an AI assistant that lets readers question its data instead of navigating the full document. That makes the report another deployment of the technology it describes, but the evidence presented remains company-supplied.