An intelligent marketplace in Iran, 2026

Everyone can ship an online shop with colour and price filters. What got hard in 2026 is the marketplace: many sellers, many buyers, uneven goods, and a reasonable lack of trust. Car listings, industrial parts, specialist services — all of them are many-to-many. If you only build a pretty catalogue, by week three the real conversation has left the system and you are left with stock arguments.

When I approach these projects I ask first: is your value in displaying goods, or in getting the right party to the right party? The answer changes search and order state.

Search needs a reason

Nobody has patience for ten pages of near-misses. Intelligent search means knowing a domestic water pump is not an industrial pump, and being able to say why a card appeared. If AI only shows “similar things” with no explanation, trust never arrives. Service matching is the same: skill, city, capacity — not just a keyword.

Do not shrug at fraud and duplicate ads. In C2C markets that is where the product grows up or becomes a spam wall. A simple rule beats a clever model with no data. Bring the model when you have a trail.

Trust is a feature, not an About page

Order state, dispute handling, and seller identity belong in the flow. You will not kill off-platform messaging. You can make the inside path useful enough that leaving is costly. If payment and settlement are in scope, they must share order state from day one, not as a month-six add-on.

Multilingual here is not a translated button. If you hope for a buyer abroad or an Arabic-speaking seller, take that locale’s route and SEO seriously — or honestly skip it. I wrote trilingual products for that mess.

From vitrine to system

Most marketplaces should start with a small loop: one category, one city, one deal type. Put intelligence on that loop. I connected this to agents in agent-native web systems and to approval in human-in-the-loop automation.

If your marketplace idea has already outgrown a simple shop, fence it in an order. “The next giant mall” is not a scope. One clear category is.