GET A QUOTE10M+
Looks
styled
outfit combinations and visual curations created by the community
Case studyEnterprise AI & Social Commerce
Building an AI-driven social commerce platform.
The Fashmates case study
Enable people to assemble a look, and then help them buy it.
At Team Tweaks, We helped to work on the platform linking a style community with product discovery, visual search, and retailer styling tools – from a clipped image to a shoppable outfit.
- 10M+Looks styled
- 200M+Social reach
- 500K+Stylists & shoppers
- iOS
- Android
- Web
- Web clipper
- B2B widgets
Historical engagement · Visuals from the 2020 business presentation
DISCOVER.STYLE.SHOP.
ONE ITEM. A COMPLETE LOOK. 
Styling + product discovery↗
- Client
- Fashmates Inc.California, United States
- Product
- Fashion & social commerceCommunity platform + Styling Cloud
- Platforms
- iOS, Android & WebWeb clipper + retailer embeds
- Engineering focus
- Discovery, AI & cloudCatalogs, content and recommendations
01 / Platform scale
A community that made fashion discoverable.
The business presentation gives a useful sense of the platform’s scale. These figures are reported by Fashmates as historical, rather than measured results resulting specifically from the development engagement.
Source: Fashmates for Business, June 2020, page 3. “10M+” refers to looks styled, not a verified SKU count.
200M+
Social
reach
reported across the platform’s connected social ecosystem
500K+
Stylists &
shoppers
creators and people discovering their next look every day
20M+
Monthly
curations
reported styling and discovery activity across web and mobile
500+
Affiliated
brands
retail partners whose products could be styled, shared and shopped
5
Platforms
delivered
iOS, Android, web, a universal web clipper and B2B styling widgets
02 / The client & the challenge
Fashion starts with “What goes with this?”
A product catalog can only tell you what is available. Fashmates is about inspiring you to see what could work together.
When Polyvore shut down, outfit creators were left without a place to collect products, assemble looks, and share ideas. Fashmates fills that gap by bringing creative tools into a community-led shopping experience.
For a team looking at Polyvore alternative app development, the big challenge is linking the canvas to the catalog. A product clipped from a retailer’s site needs usable imagery, meaningful attributes, and a route back to the item being sold.
- 2018 Polyvore closes
Millions of outfit creators lose their favourite styling canvas.
- The gap Fashmates steps in
Community styling meets AI-driven product discovery.
- Feb 2022 Brand Capital invests
Strategic investment to support expansion into India. Read the announcement ↗
Three problems to solve at once
Hover or tap a challenge to open it.
Aggregate information from multiple catalogs
Vendor product feeds and clipped products arrived with different images, descriptions and prices. There was a need for a standard record that could be searched and reused.
- Feeds from 500+ affiliated retail brands
- Products clipped from almost any store
- One consistent, reusable product record
Make images useful to search.
A neckline, fabric, or silhouette could be more meaningful than a product name. Computer vision and language processing made it possible to add structure to that information.
- Computer vision auto-tagging
- NLP attribute extraction from descriptions
- Visual similarity search
Take a look beyond the canvas.
The same styling will work in the community, on a merchant's product page and in a shopper's next email.
- Community canvas and social feeds
- Retailer product and listing pages
- Re-engagement and abandoned-cart emails
B2CFor creators & shoppers
A place to collect, create and share.
Web clipping, a drag-and-drop canvas, personal inventories and shoppable outfit sets formed the community side of the platform.
- Web clipper
- Outfit canvas
- Social discovery
B2BFor brands & retailers
A styling layer for the storefront.
The B2B Styling Cloud brought look galleries, visual recommendations and personalised curations into a retailer’s own shopping journey.
- Embedded widgets
- StyleProfile
- Complete the Look
03 / Fashmates app architecture
From a product image to a shoppable look.
The project brief describes a decoupled AWS architecture: ingest product data, enrich it, index it, and make it available for the community features and for retailer features.
AWS cloud / EC2 / S3 / Lambda / Microservices
03Index & serveSearch · widgets · community
Knitwear Long sleeve Cream02Process & enrichVision · NLP · cutout
Feeds Clipper Apps Canvas
01Ingest & createFeeds · clipper · apps
Ingest & create
Products enter the platform.
From the retailer feeds, through the JavaScript web clipper, the native apps and the web canvas editor: they all funnel product information in through HTTP APIs.
- Retailer and affiliate feeds
- JavaScript web clipper
- Native iOS and Android apps
- Web canvas editor
Output → HTTP APIs
Process & enrich
Images become useful data.
Computer vision and language read each product, crop the image and attach attributes that can be used for search and styling.
- Computer vision auto-tagging
- NLP attribute processing
- Image cutout service
- Visual similarity processing
Output → Enriched product records
Index & serve
Discovery reaches the shopper.
Enriched records are indexed for fast search and served to the community, retailer widgets and recommendations.
- Elasticsearch product index
- PostgreSQL user and set data
- Styling Cloud API / widget CDN
- Retailer and community views
Output → Search + recommendations
Cross-cutting services
- Privacy & moderation
- Affiliate synchronisation
- Retailer embed delivery
Technology stack
- Computer vision
- NLP
- Elasticsearch
- PostgreSQL
- AWS EC2
- S3
- Lambda
Collection is not to be confused with discovery
Clipping and feed ingestion is a separate process than search. By processing product images and descriptions before indexing, the results can be fed into filters, visual discovery, and outfit creation.
Reuse of this kind of content where shoppers are.
The community platform and the B2B widgets relied on the same type of product and stylistic information. A look could be created by a creator on the canvas, then appear in a retailers’ gallery or recommendation panel
Explore our e-commerce app development solutionsLogical data flow reconstructed from the supplied engineering brief — an overview, not a verified production deployment map.
THE OUTFIT CANVASPRODUCT CONCEPT
Collect → Arrange → Tag → Share4 pieces. One outfit.
Do not just search by a phrase, but by resemblance.
Visual similarity provided a new way to discover products, giving users a way to explore items that looked like the one they were currently looking at, in addition to text search and attribute filtering.
05 / Security & moderation
A creative community
requires close oversight
Interaction and dialogue are integral to the product offering. Ensuring the safety and security of these interactions is a design responsibility
Within the technical specification are requirements for redaction and content moderation based on Pangea's technology. A distinct Pangea customer case study details how Fashmates utilize redaction in the context of offensive language and profanity.
500K+registered stylists and global shoppers protected by API-based PII redaction
Read the Pangea customer story Personal information
This documented security scope aims to detect and redact confidential information such as personally identifiable information and payment-card information.
From the engineering briefPangea Cloud API redaction protected user privacy and credentials in real time across 500,000+ registered stylists and global shoppers.
Community conversations
Moderation deals with offensive language and curse words in user-generated content and chat.
Our solution continuously monitors user chats and public comments to identify profanity, abusive language and PII leakage in real-time within a global social community
Content before publication
The brief covers the process of filtering content before it gets posted to public feeds. These mitigations decrease the chances of exposure, but cannot ensure that every objectionable submission is identified.
From the engineering briefUser-generated social content, comments and non-compliant images were filtered before publishing to public feeds.
Official verification anchor
Business verification is conducted by referencing public company directories and observing coverage in the global business press, ensuring the platform’s history is verifiable.
See the referencesTimes of India investment report, Softonic app listing and the Pangea customer case study are linked in the next section.
06 / Project references
The history behind the work.
The business presentation describes the product offering. These outside sources provide context about the company, its mobile presence, and the security integration.
3external sources
you can open and check
- Investment announcementThe Times of India Brand Capital International’s February 2022 strategic-investment announcement.Mediawire · Content produced by the investor
- Mobile product recordAndroid app listing A third-party directory record describing Fashmates’ outfit creation and community features.Softonic · Platform reference, not an endorsement
- Security integrationPangea customer story A vendor reference describing Fashmates’ approach to redaction in community conversations.Pangea · Customer case study
This page describes a historical engagement. The references support specific company or product facts; they are not independent verification of every architecture detail or Team Tweaks’ delivery scope.
07 / Your next commerce product
What’s the next thing shoppers should discover?
Visual search, catalog tagging or a social shopping platform – all great things to have, but where do you even begin? Start by figuring out the experience you want for your shoppers. We’ll help you translate that vision into a practical technical scope.
Let’s work through
- 01
Your catalog
Product feeds, image quality, attributes and update frequency.
- 02
Your discovery experience
Search, recommendations, styling tools and community content.
- 03
Your delivery plan
Apps, cloud services, retailer integrations and launch priorities.
Partner with a mobile app development company in Chennai experienced in building software for international businesses.
