Amazon Clone Features: The Full Multi-Vendor Commerce Scope
This page describes proposed scope for a build in this category, agreed in writing per project. Nothing here is a finished product or a fixed bundle.
Every capability below is grouped by who actually touches it - the customer buying, the vendor selling and getting paid, and the delivery agent closing the loop at the door. Each is proposed scope, confirmed in writing before work starts.
Feature Set by Role
Every capability below is proposed scope, grouped by the person who uses it.
Customer
- Keyword search with autocomplete, filters and sorting
- Featured, most-demanded and best-selling rails
- Deal of the day, time-boxed flash deals, clearance storefront
- One cart spanning several vendors at once
- Guest cart tracked by session and merged on login
- Coupon, wallet and loyalty applied together at checkout
- Refund request with reason and evidence
- Restock subscriptions, wallet, loyalty and referral
Vendor
- Register with KYC, approved by an operator
- Product CRUD with variants, stock and bulk import
- SKU-level rows with their own price and quantity
- Stock thresholds with low-stock alerts
- AI-assisted descriptions, SEO titles and alt text
- Order queue and refund responses
- Sales analytics and withdrawal requests
- Web panel or Android app, same data
Delivery agent
- First-class identity with KYC and admin approval
- Assignment feed with accept and reject
- Navigation to vendor and to customer
- Live location broadcast and status checkpoints
- One-time-code verification at the door
- Cash collected written to their own balance
- Reconciled to the operator wallet on settlement
- Earnings view and withdrawal request flow
A delivery agent is a proposed identity with its own KYC, wallet and approval state, not a flag on a customer account. That distinction is what makes cash reconciliation possible at all.
Amazon vs the Proposed Build vs Building From Scratch
The same capability, three different paths. The middle column is proposed scope, agreed in writing per project - not an implemented product.
| Capability | Original Amazon | Omnistu proposed build | Build from scratch |
|---|---|---|---|
| Multi-vendor catalog | Yes - the core product | Variant matrix, SKU-level stock, unlimited category depth | The part everyone starts with |
| One cart across vendors | Yes | Yes, with shipping estimated per vendor before checkout | Splitting an order cleanly is where it gets hard |
| Your own delivery fleet | Yes - built over decades | Agent identity, assignment feed, live location, OTP handover | Ask whether the fleet is in scope or quoted as a second product |
| Cash on delivery | Yes, in some markets | Per-agent balance, operator settlement, its own transaction table | Ask how cash is recorded, whose balance it sits on, and how it reconciles before you sign |
| Four separate wallets | Not publicly detailed | Customer, vendor, delivery agent and operator, append-only | Ask how many wallets the build has and whether every movement writes a history row |
| Commission control | Set by Amazon | Global, per category, per vendor or per product, reversed on refund | Ask whether commission reverses automatically when an order is refunded |
| Payment methods | Extensive | Eleven gateways (proposed integrations) plus offline methods and COD | One gateway, then an integration per market |
| iOS apps | Yes | Not in the proposed base - Android builds proposed, iOS quoted as an add-on | Two platforms to build and submit |
| Source code ownership | N/A | Full backend and all three Flutter projects, nothing obfuscated | Full ownership - you are building it |
The iOS row is stated plainly because it is the thing buyers most often assume. Three Android builds are proposed base scope, recorded in the supplied specification as Android builds; iOS builds - App Store signing, submission and review cycle - are proposed scoped work, subject to validation in your own testing.
How It Works, End to End
One order, followed from browse to settled money - including the doorstep, which is where most marketplace platforms stop. Each step below is proposed flow, agreed in writing per project.
1 - Browse and add to cart
Keyword search with autocomplete, filters on price, brand, rating, vendor and attributes, and merchandising rails for featured, best-selling, deal of the day and clearance. One cart spans several vendors at once, with shipping estimated per vendor before checkout, and a guest cart is tracked by session and merged on login.
2 - Checkout, however they want to pay
Coupon, wallet and loyalty apply together, scoped to product, category or cart with per-user usage limits. Gateways, operator-defined offline methods and cash on delivery all terminate in the same order. Placement is guarded by an idempotency check, so a double tap on a slow connection does not create a second order.
3 - Split, assign and dispatch
The order splits across vendors, stock decrements at SKU level, and each line carries its own delivery status so a multi-vendor order can be partially fulfilled. A delivery agent picks the job off an assignment feed, navigates to the vendor and then the customer, and broadcasts live location through the checkpoints.
4 - Handover, verified
A one-time code is verified at the door rather than the agent simply marking it delivered. On a cash order, the amount collected is written to that agent's own balance and tracked in a dedicated transaction table rather than being treated as an ordinary payment.
5 - Settle, or reverse
Commission is deducted from the vendor wallet when the order completes and reversed automatically if it is refunded, with loyalty points reversed on cancellation. Cash held by the agent reconciles to the operator wallet on settlement, and every movement writes an append-only history row so any balance can be re-derived.
The step to interrogate
Step five is the one to test in any platform you evaluate. A refund that reverses the payment but not the commission, or cash that never reconciles, leaves a balance nobody can explain - and you will find it at year end, not at launch.
Every Feature Earns Its Place
Nothing on the proposed list is there because it demos well. Each capability exists because leaving it out breaks something specific and expensive.
| Capability | Why it is not optional |
|---|---|
| Four separate wallets | Because a customer refund, a vendor payout, an agent's cash and your commission are four different liabilities. One balance column means you cannot answer who is owed what. |
| Append-only wallet history | Because a balance you can only read is a balance you cannot defend. Every movement writes a row, so any figure can be re-derived from the ledger. |
| Cash in its own transaction table | Because treating cash as an ordinary payment hides the gap between money collected at a door and money in your account. |
| OTP at handover | Because "marked delivered" is not proof, and disputes over undelivered cash orders are the most expensive kind. |
| Idempotency on order placement | Because a double tap on a poor connection should not charge twice or reserve stock twice. |
| Per-line delivery status | Because an order split across three vendors will not arrive at once, and forcing one status on it makes the tracking page lie. |
| Commission reversed on refund | Because a refund that returns the customer's money but keeps your commission is a vendor dispute waiting to happen. |
| Field-level edit history | Because when a price or a status changes, the question is always who changed it and when. |
| Zip allowlists | Because offering checkout where you cannot actually deliver converts a sale into a refund and a bad review. |
The Technology Behind the Proposed Scope
A conventional, hireable stack, deliberately. Every element below is the proposed architecture, confirmed in writing per project - the point is that the developers who maintain it after handover should already know it.
Backend
Proposed: PHP 8.2 on Laravel 12 with nWidart modules, running controller to service to repository to model with transactions at the service boundary. Modules for AI, blog and tax enable and disable without touching core, and observers, events and listeners drive notifications and ledger writes. Versioned REST APIs serve every client.
Mobile
Proposed: Flutter with Dio and Provider over GetIt - customer, vendor and delivery builds from one shared architecture, Sanctum bearer tokens carrying role claims, Firebase Cloud Messaging per app and Google Maps in the delivery build.
Data
Proposed: MySQL 8.x and Redis 7.x with InnoDB and utf8mb4 throughout, soft deletes on customer, vendor, agent and product rows, composite indexes on the hot paths, and append-only wallet histories.
Scale of the schema
Proposed: 114 Eloquent models over 131 normalized tables - a count from the specification inventory, confirmed per project. Nothing exotic, and that is the point.
What Is Not Included in the Proposed Base
These items are additive scope beyond the proposed base, quoted in writing per project - labelled configuration-required or integration-required rather than implied away. None of them is implemented or verified by this preview.
iOS applications
Three Android builds are proposed base scope, as recorded in the supplied specification. iOS builds are proposed scoped work with signing and submission, subject to validation in your own testing.
Gateway credentials
Gateway integrations are proposed scope; each one needs your own merchant account and approval.
SMS and email provider
Integration-required - your accounts, your sender reputation.
Firebase project
Push notifications run through your own Firebase project.
AI provider key
The AI module generates descriptions, SEO copy and alt text against your key, with per-feature token quotas and a cost log per call proposed.
Catalog data
Products, imagery and vendor recruitment are the business, and they are yours.
Tax rates
Tax classes binding per zone are proposed scope, but the rates for your jurisdictions are yours to set.
ISO 27001 and SOC 2
This preview supplies no ISO 27001 or SOC 2 evidence; no certification is claimed here. Any compliance evidence gathering is separate work, scoped per project.
Currency exchange rates are proposed as operator-set rather than fetched automatically - deliberate, because a marketplace that silently reprices a catalog against a live feed surprises vendors and customers at the same time.
Frequently Asked Questions
How does the delivery fleet work?
As proposed scope: delivery agents are a first-class identity with KYC and admin approval, not a flag on a customer account. They get an assignment feed with accept and reject, navigation to vendor and customer, live location broadcast, status checkpoints, one-time-code verification at the door, and a wallet with a withdrawal request flow.
Is cash on delivery properly handled?
In the proposed design, cash collected is written to the delivery agent's own balance, reconciled to the operator wallet on settlement, and tracked in a dedicated transaction table rather than being treated as an ordinary payment. It is one of the places general commerce scripts give up.
Can vendors manage everything themselves?
That is the proposed flow: vendors register with KYC, get approved by an operator, then run their own shop - product CRUD with variants and stock, bulk import and export, AI-assisted descriptions, an order queue, refund responses, sales analytics and a withdrawal request flow, from either the web panel or the Android app.
Which payment gateways are supported?
The specification covers eleven: Stripe, PayPal, Razorpay, Mercado Pago, Iyzico, PhonePe, Paytm, PayU, SSLCommerz, Flutterwave and PayTabs, plus operator-defined offline methods and cash on delivery. All are proposed integrations, and each needs your own merchant credentials - which is why they are labelled integration-required.
Are iOS apps included?
No. The three Android builds are proposed base scope, as recorded in the supplied specification. iOS builds - signing, submission and the review cycle - are proposed scoped work, subject to validation in your own testing.
What does the AI module actually do?
Proposed: it generates product descriptions, SEO titles and descriptions, tag suggestions and image alt text through OpenAI, with per-feature token quotas and a log recording tokens and cost for every call. It needs your own key, so it is integration-required rather than included.
Independence and third-party names
Omnistu is an independent software development project. We are not affiliated with, connected to, sponsored by, or endorsed by Amazon. "Amazon clone" is used descriptively - it is how the software industry refers to building a platform with similar functionality, and how buyers search for it. Amazon and all other third-party names and marks are the property of their respective owners, referenced here solely to describe the category of software offered.
Discuss your scope by message
Tell us your commerce categories, your expected vendor count and your delivery model. We reply by email with a scoped proposal - no calls required, no phone numbers collected.