Mazzaneh: The Architecture of Demand-Driven Commerce

The first version of Mazzaneh answered a narrow question: what if buyers could describe what they wanted, and the right sellers came back with private offers, instead of buyers chasing the right seller through search and category trees? That answer became Mazzaneh Begir — one request, multiple private quotes, best match. It worked. It also exposed a second-order problem the original idea did not see: a request-based system collapses if the rest of the platform feels empty while the network responds.

This article is about what happened after that realization. Not features for their own sake, but a set of design decisions that, together, formed a coherent architecture for a kind of commerce platform that did not exist before: one where the seller's effort approaches zero, the buyer's wait time approaches zero, the advertiser's budget waste approaches zero, and the user is the system's strategic asset rather than its product.

1. The Visual Layer: Mazzaneh Gram

As Mazzaneh grew, the same friction kept appearing in seller onboarding interviews. Many sellers did not have products ready to display. Many did not want to upload listings at all. In wholesale, in traditional retail, in service businesses, content creation was not a feature — it was a wall.

A request-based system has a parallel problem on the buyer side. Some sellers respond slowly. Buyers will not sit inside an empty app waiting. We needed a bridge — something that kept buyers engaged while the network worked, without forcing sellers into the kind of heavy content creation that had stalled adoption everywhere else.

Mazzaneh Gram is a visual showcase for businesses — photos, descriptions, prices, organized into clean albums. Buyers browse fast. Sellers showcase without building an e‑commerce system.

The design choice that mattered most was permission to import existing Instagram content directly into Mazzaneh. Sellers had years of inventory already photographed; rebuilding from scratch was the wrong ask. Treating their existing content as a first-class import path turned a multi-week onboarding into a one-day onboarding. More on Mazzaneh Gram.

Categorized Albums

In early adoption, speed of relevance matters more than completeness of inventory. A buyer who knows what they want should not have to scroll past unrelated posts to find it. Albums solved this. A real estate agent could separate listings into sales, rentals, and leases. A musical instrument store could separate strings from percussion from accessories. The buyer's cognitive load dropped, the seller's organizational signal went up. More on Albums.

2. The Zero-Effort Storefront

For one segment of sellers — traditional, wholesale, B2B — even Gram was too much. They did not want to upload anything. They did not want to take photos. They did not want to manage social media or operate 24/7 online support. They wanted to receive requests and respond. That was the whole job.

If the platform required listings from this segment, the storefront looked empty and trust collapsed at exactly the moment we were trying to earn it. So we inverted the model. Sellers would not need to build content. We would build a professional storefront for them automatically, using the basic information collected during a quick onboarding call.

Two design moves made this work. First, we displayed relevant product categories on each storefront page using visual designs that looked like actual products, so the page never felt empty even before any real inventory existed. Second, we added a small action icon below each category that allowed any buyer to request a private quote (Mazzaneh Begir) for that specific category, instantly. The seller never had to upload anything. The buyer never had to wait through an empty page. More on Storefront Profiles.

3. Multi-Channel Delivery

A request-based marketplace fails if delivery depends on a single channel. Some sellers have the app installed but miss notifications. Some uninstall and reinstall under varying connectivity. Some prefer SMS. Some use WhatsApp as their de-facto business inbox. In real-world conditions, no single channel can be treated as critical.

So we built request delivery as a fallback protocol rather than a single send. The system attempts in-app notification first. If the seller does not confirm receipt within a tight window, the request falls back to SMS. If that also fails the response check, it falls back to WhatsApp (linked to the seller's verified profile during onboarding). From the buyer's perspective, the experience is unified — replies arrive inside the app in a single conversational flow, regardless of which channel actually carried the request. More on Multi-Channel Delivery.

4. Why Wholesale Sellers Adopt the Model

Three properties of the architecture above, taken together, made Mazzaneh structurally well-fit for wholesale and traditional commerce in a way no listings-first marketplace can be:

5. Discovery Through the Map

To prevent the platform from feeling empty in a new region, the system needed strong inventory signals fast. Importing content from Instagram was the first piece. Auto-generated category pages were the second. The third was a geographic surface — the map.

As categories filled, buyers could browse, request a quote, or open a conversation through storefront-style layouts inspired by familiar shopping experiences. To reinforce credibility at neighborhood scale, businesses appeared as map pins. A user could open the map, see who was nearby, open a store profile, and start chatting instantly. Discovery became a local, visual experience instead of an abstract search query. More on Map Discovery.

6. From Marketplace to Daily Assistant: The Path to Zoyan

Everything described above answers the same architectural question: how do you operate a demand-triggered commerce platform that does not collapse on either side? The answer is real, and at a certain point the platform crosses a threshold where everyday economic activity flows through it.

But a real marketplace was never the end state. From the earliest design notes, Mazzaneh was meant to evolve into something that integrates with daily life directly: a 24-hour intelligent assistant and consultant. We called it Zoyan.

Reaching Zoyan required reverse-engineering from the destination. What does a personal assistant need to know about you to be genuinely useful, not just responsive? What signals about your preferences, your routines, your purchasing patterns, your skills, your interests cannot be reliably collected anywhere else? Those signals had to come from somewhere, and they had to come with consent — both legally and structurally — because anything less would not survive contact with the regulatory environment of the markets we wanted to enter.

7. Zero-Waste Advertising

A simple observation made the next step obvious. Sellers are constantly paying to reach customers. A large fraction of every advertising budget is burned on the wrong audience — impressions that reach people who will never buy, clicks from outside the relevant geography, follows that never convert.

A music store should not pay to reach people uninterested in instruments. A pet shop should not pay for impressions outside the population of pet owners. The same budget, flowing only to the right audience, should produce orders of magnitude more conversion at lower cost.

This became the core concept we called Zero Waste Advertising. The advertiser spends less. Followers are precise. Budget is not burned on irrelevant people. And — critically — when a target customer follows a store inside Mazzaneh, the configured advertising amount is deducted from the seller's account and credited to the user's account. The exchange is direct: attention becomes an accountable, reward-based channel. The user earns. The seller pays only for relevant attention. More on Zero Waste Advertising.

8. Unlocking User Data — Without the Surveillance Trap

For Zero Waste Advertising to function, the platform needs to know who the right audience is for which store. That means user signals: interests, traits, tastes, skills. These signals define which businesses each person is a target customer for.

Most platforms answer this question through surveillance — inferring user attributes from behavior, often invisibly, often beyond the user's explicit awareness. The economics work, but the trust does not. We answered it differently. Users tell us who they are in a structured way, voluntarily, in exchange for income. The data layer is consent-first by construction.

The critical privacy property: sellers never see user identities. They see only relevant followers indexed by anonymous user codes (for example, user #2123). The advertiser's goal in this model is not to identify the individual buyer — it is to attract a target audience that recognizes the store and sees its products. The information asymmetry that classical advertising depends on is preserved, but the identification it usually requires is not. More on User Income and Targeting.

9. Mazzaneh Board: Gamified Product Education

Once user-income exists, the next design question is what users actually do to earn it. Following stores is one channel. But following is a shallow signal — it does not transfer real product knowledge from advertiser to audience.

So we added a second channel. Mazzaneh Board shows a product advertisement, together with its full description and features, to the target audience. The user is asked to read and remember the content. After that, they participate in a 20-second quiz with four multiple-choice questions. If they answer all correctly, they win a reward.

The mechanics matter. A billboard cannot convey the full message of a product. A 30-second pre-roll cannot transfer technical specifications. A static banner cannot test whether the audience absorbed what they saw. The Board structure does all three, and it does them with a verification step — the quiz — that filters genuine attention from passive impression. More on Mazzaneh Board.

10. Verification: The Layer That Holds the System Honest

Once user-income and targeted advertising both exist on a single platform, verification becomes non-negotiable. Without it, anyone could claim false traits and become a "target audience" for every business. The economics would collapse within weeks. So we designed a verification path that uses time, limits, and identity checks at withdrawal to keep the system fair.

1. Throttled participation.To prevent users from reaching withdrawal thresholds too quickly — and to maintain professional standards while ensuring daily engagement remains real — participation in ads and store follows is capped per user, per period.
2. Identity verification at withdrawal.When a user withdraws funds from their Mazzaneh account, real identity information is required (national ID or equivalent). This prevents the same person from operating multiple accounts under different usernames and phone numbers to multiply income.
3. False-information enforcement.If a user declares false attributes and the system later discovers this, the entire account is blocked and the balance becomes non-withdrawable. The associated IP is barred from re-entering the platform. Funds paid to that user from advertisers — for follows or quiz rewards delivered against false profiles — are returned to the advertiser's account.

The verification path includes one additional check. For each declared characteristic, the user must complete at least one purchase from a relevant business within a maximum of three to four months. This converts declared preferences into behaviorally validated preferences over a defined window, and it does so without requiring the user to surrender their identity to the advertisers themselves. More on Verification and Trust.

The Shape of What Was Built

Read from one end to the other, the architecture above describes a system in which several normally-conflicting requirements coexist. Sellers do not have to build content, but their storefronts are visually credible. Buyers do not have to search, but discovery still works. Advertisers spend less, but reach more of the right people. Users contribute data, but only data they choose to declare and only for compensation. The platform mediates, but does not act as a data broker.

Each of these properties is moderately interesting in isolation. The argument of this article is that the value emerges from their combination — from the way one design choice becomes the precondition for the next, until the result is an integrated platform whose properties no individual feature could deliver alone. That integration is also why the platform was always intended to extend into Zoyan: once consent-based, behaviorally-validated user understanding exists at scale, a personal AI assistant has the information layer it needs to be genuinely useful rather than generically responsive.

The platform was not the destination. The platform is the data and trust architecture that makes a 24/7 AI assistant — Zoyan — possible.

That is the system we built, the reasons each layer exists, and the direction the whole structure points toward. The next stage is partnership: the integration of this architecture with frontier AI infrastructure, and the deployment of Zoyan as the consumer-facing surface of an assistant that knows the user well enough to be useful and respects them enough to be trusted. We are open to that conversation.