Flow of Presentation

We will be looking into use of AI by Flipkart in the following:

Introduction
Initial Implementation of AI
Building Customer Search Experience
Customer Recommendation Engine
MIRA- The AI bot
Launch of the brand- Billion
Integration of AI to Myntra
ARTIE-The AI Fashion bot
Inventory Management
Sortation Center
Automation in Warehouse
Assistance to Sellers
Security in Supply Chain
Quality Check
Transaction
Delivery
Introduction
 Flipkart Private Limited, is an Indian e-commerce company based in Bengaluru, India. It was
founded by Sachin Bansal and Binny Bansal in 2007. The company initially focused on book
sales, before expanding into other product categories such as consumer electronics, fashion, and
lifestyle products.

 The idea to integrate AI, came because of its focus on problems such as personalisation, product
discovery, address understanding, search, and fraud detection

 Integrating AI has helped reducing costs, and reducing friction in the delivery supply chain

 In the next five years, Flipkart's vision is to harness AI to solve complex problems unique to
India and to that end it has carved out an internal unit called AIforIndia
Initial Implementation of AI
 In 2015, Flipkart had acquired the F7 Labs in the Silicon Valley to focus on AI and
machine learning

1. The 10-person Labs team is working on creating a model to determine which products
get returned the most, so as to build a more cost effective way to manage those returns

2. Has automated the moderation of the comments and product reviews that appear on
the site

 In 2017, after partnering with Microsoft, Flipkart leveraged AI, machine learning and
analytics capabilities in Azure, such as:

1. Cortana Intelligence Suite

2. Power BI

 This helped to optimize its data for innovative merchandising, advertising, marketing
and customer service
Building customer search experience
The firm has created at least 40 insights for each of its over 100 million customers — from
GPS locations and page scrolls, to preferred brands and frequency of purchase — which
has helped to dissect customers buying psyche, boost brand awareness and detect fraudulent
buyers

The AI software use this large amount of data for the functioning of the customer
recommendation engine, it takes advantage of-

1. Image recognition — a deep learning application — for suggestions, product discovery,

product grouping and catalogue search

2. Conversational Search Experience provided by the AI bot MIRA

Building customer search experience

 On low mobile bandwidth networks, artificial intelligence-enabled

algorithms narrow down product searches to a smaller number of pro
ducts without sacrificing image quality, discovering more personalised
products that meet the relevant aspects required by the buyer

 The majority of products detected on low mobile bandwidth are from l

ocal sellers who cater to hyperlocal taste
Building customer search experience

 Flipkart Launched Assistant Saathi To Enhance Shopping Experience

 Flipkart Saathi comes after Flipkart acquired artificial intelligence company Liv.ai, in August 2019
 Flipkart Saathi will be available in text and voice interface
 Flipkart Saathi targets users from Tier 2, Tier 3 and beyond
 Flipkart launched a Hindi interface in September 2019
 The feature has been specially designed keeping in mind the users from Tier 2, Tier 3 cities and
rural regions, who are getting on board as customers rapidly
 It help users navigate through the website with the help of audio and textual instructions
Customer Recommendation Engine
 This AI enabled tool, use bots which helps in Marketing the right product to the right target
group.
 Bots like Mira and Artie are enabling this

The AI tool works in the following way:

 According to what customer has ordered in the past, related to that product the
customer is recommended product.
 Suppose two people buy same product, then if one of the person buy some X
product, the AI will predict and recommend the other customer for the same X
Customer product, feeling the choices might be same as in the past they have bought same
Oriented product.

 For every user, preference is made by similarity score.

 Similarity score is given by comparing customers who have ordered
similar stuff in the past.
 Higher the similarity score with respect to a user, recommendation
of what that user is buying, is given to the 2nd user.

 Retailer claims to take advantage of image recognition, speech recognition, text

recognition, as well as personalization, search and recommendations to increase
their sales.
 Forecasting of demand has allowed retailers to deliver at the earliest
MIRA: The AI BOT
Mira guides shoppers with relevant
questions, conversational filters,
shopping ideas, offers and trending Using sequence-to-sequence
collections. deep-learning framework that, given a
customer query, provides up to five
The machine learning algorithms can possible responses
also detect incorrect images and
morphed images. With Project Mira only in an infancy
Posting the right images on the stage, Flipkart is seeing nearly a 12%
platform is crucial to getting customers cart addition and a dramatic
interested in a product improvement of seller experience.
Sellers intentionally or unintentionally
post duplicate products, which The bot enables estimating if deliver,
increases user effort in scanning to the say, in (an optimal) two days is a
desired set of products possibility or not, because if it takes
longer customers fall off, they cancel
midway
This AI bot ensures the descriptors of a
product like colour, pattern, etc., are
accurate
MIRA
 This goal of Mira is to scale the in-store experience of having a sales associate – but via
artificial intelligence and through digital channels. Users with broad intent (searching for, say,
shoes or bedsheets) are guided with relevant questions, conversational filters, shopping ideas,
offers and trending collections. This technology is already in use at Flipkart.

 For Example:
Say, someone comes to Flipkart searching for an air-conditioner. Because of Project Mira,
Flipkart now asks buyers about what kind of AC they want, the tonnage, room size, brand,
and such. It is a beginning in helping customers find the exact product they need in online
settings that aren’t exactly easy to navigate.

 Like in offline shopping where the sales executive asks a customer about her size and other
requirements before making a suggestion, Flipkart is trying to get AI to do something similar.
Launching the brand- Billion
With the help of tools like MIRA, Flipkart was able to track customer likings, and acc
ordingly launched its own brand, Billion, that would satisfy the interests of the custo
mers.

 Billion is one of Flipkart’s brands shaped by extensive review analysis using

artificial intelligence tools.

 Billion air-conditioners and more products like backpacks with more storage space,
fast-charging power banks, extra-absorbent diapers and kitchen pans with thicker non-stick
coating propelled Billion to popularity among shoppers on Flipkart

 The firm claims that returns on Billion products are 50% less compared to all other brands
sold in the marketplace.
Integrating AI to Myntra

 Flipkart acquired Myntra in 2014, and in 2015 began integrating it with AI

 Artificial intelligence solutions have reduced design time for Myntra’s

innovation lab — named Rapid — from six months to just one

 The Rapid team at Myntra, which oversees fashion brands like Moda Rapido and
Here & Now, has more data scientists than it does designers

 They’re very, quick on supply chain. It takes seven to 45 days for a collection to drop.
One of the t-shirt platform’s turnaround time is 24 hours, and designs drop in even a
few hours sometimes
 Intelligent Fast Fashion

 The AI system Ratatouille is an Artificial Intelligence-based software us

ed by the Myntra’s in-house Rapid team

 Ratatouille tells just what designs are sure to be top-sellers

 This software is more professionally known as Vorta.

 Vorta is well on its way to becoming a global Software as a Service (Sa

aS) product and Flipkart has already signed up with one of the
world’s leading apparel brands and few of India’s top western and et
hnic wear brands on the Vorta platform
ARTIE – The AI Fashion BOT

Flipkart's fashion unit Myntra, use an

indigenously developed smart bot called
as Artie, that gathers consumer insights
to create a range of on-trend, fast fashion
apparel under the 'ModaRapido' brand.
 Inventory Management

THE BUTLER SYSTEM THE SORTER SYSTEM

The Butler system is an automated inventory The Sorter is an advanced robotic sortation
storage, replacement, and order picking facility system that has automated the sorting processes
that is deployed at warehouses, fulfillment centers in warehouses and fulfillment centers. Two
and distribution centers. It consists of an double-deck and five single-deck sorters of
advanced autonomous mobile robot called Butler; capacity 6,000 sorts/hour each, apart from four
Mobile Storage Units (MSU) or racks; pick-put sorters of capacity 3,000 sorts/hour each were
stations; and Smart Charging Stations. installed at Flipkart’s Hyderabad fulfillment
center and eight transport centers...
Sortation Center

At the Flipkart sortation center tucked away in the outskirts of Bengaluru, humans
and AI-powered bots — known as Automated Guided Vehicles (AGVs) — work
seamlessly together to achieve over 4,500 shipments an hour at twice the speed
and with 99.9% accuracy.

These AI bots are loaded with sensors, which means that, despite there being
over a hundred of them in an enclosed area, they don’t bump into each other

 They run 24/7, except when they automatically guide themselves towards
designated charging points when batteries run low

The process can be seen here: https://youtu.be/p8RlPfKk0EE

Automation in Warehouse

 Warehouses are of 2.2 lakh share feet, with storage capacity of 5.9 lakh cubic feet, and
they handle over million products
 Each item goes through 6 quality checks
 Measured and stamped using barcodes, which contain all data related to packaging as
well
 Products are moved using battery operated vehicles, while Forklifts are used for
pre-designated locations
 As the user places an order, AI locates the nearest warehouse and reserves the product in
that particular warehouse
 Products move on automated conveyer belt, which handles 5000 packages in an hour
 These conveyer belts are equipped with AI enabled robotic arms, with IOT sensors,
which use dynamic sorting logic to sort items based on PIN codes
Assistance to Sellers
1. Machine learning applications also support sellers on Flip
kart’s platform. This makes it easy for shop owners t
o calculate demand or revenue per impression (RPI) for a
product, as well as develop price discovery of a specific
product.

2. The questions artificial intelligence can answer go as dee

p as “What are the demographics of a person who
wears a mandarin collar?”, to “What colour works for
which demographic?” and even, “What fabrics see t
he most returns?”
Security in Supply Chain

Intent Modelling Fraud detection

Intent modelling is done by AI software, Suppose a phone is bought by a customer, and the
which ensures the products are bought n he/ she claims that they got soap bar. Now to the who
by genuine buyers and not resellers. le. value chain, whosoever are the participants
. are, AI checks their past activity and accordingly score
s them. That is how they detect fraud, and block sell
ers with minimum score.
Quality Check
 In another initiative named Flipkart Utkarsh (which means ‘excellence’,
the e-commerce firm recently harnessed artificial intelligence to improve
on the quality of products sold by its 100,000 registered sellers

 Artificial intelligence and machine learning models derive insights from

the data collected from buyers and sellers, which are then used to create
guidelines to improve quality and reduce returns

 Natural Language Processing (NLP) is another application that Flipkart i

s leveraging to help to moderate or clean up the reviews section partic
ularly during big sale. days, like The Big Billion Day, flagging off delib
erate negative reviews
Transaction

Suppose there is an area where there are lot of cancellations, or rep

ort of payment issues at the time of delivery, then with adding variables like
average income, average age, AI enable or disable COD option.
Delivery: Ekart Services
The AI system figures out what products customers are likely to buy, and anti-fraud algorithms
help in determining whether the order is genuine
In the absence of a formal geo-spatial classification, addresses in India are organized by Postal
Index Number ( PIN codes). Factors like lower literacy rates lead to customers entering incorrect
PIN codes, or commit spelling errors while filling up addresses

Data Science team at Flipkart had to understand how this unique knowledge could be converted
into a ML model. The team started with approximating a mental model employed by the sorting
personnel to classify addresses correctly, and then built a ML model that would attempt to learn
incrementally from these labels
Flipkart, artificial intelligence has helped the firm classify and identify addresses with 98%
accuracy, reducing delivery time by at least three hours per delivery hub

Software even maps the shortest and least complicated route to the customer’s doorstep

It accordingly assigns the shipment to a field agent to deliver it to home by routing the package
from a warehouse to the destination city
Thank you