SME Times News Bureau | 27 Aug, 2023
In an exclusive interview with Vinod K. Singh, CTO for
Concirrus Ltd, a reputable insurance software business with headquarters in the
UK, said that Blockchain technology offers a transformative solution by
enhancing transparency, authentication, and security in international
Excerpts of the interview…
1. Cross-border e-commerce involves
complex logistics and regulatory challenges. Could you discuss how blockchain
technology can be employed to address issues related to transparency,
authentication, and security in international transactions?
Answer from Vinod
Certainly! Cross-border e-commerce does indeed present
intricate logistical and regulatory hurdles. Blockchain technology offers a
transformative solution by enhancing transparency, authentication, and security
in international transactions.
Blockchain's decentralised ledger records every step of a transaction, ensuring
an immutable and transparent trail. This transparency minimises the risk of
fraud, counterfeiting, and disputes, as all parties have access to the same
Blockchain enables robust identity verification through cryptographic
techniques. This ensures that both buyers and sellers are genuine, reducing the
likelihood of fraudulent transactions and protecting the integrity of the
Blockchain's encryption and consensus mechanisms enhance data security. It
mitigates data breaches and cyberattacks by eliminating a single point of
failure and distributing data across the network, making it exceedingly
challenging for malicious actors to compromise the system.
Blockchain's smart contracts automate and enforce agreements, streamlining
cross-border transactions. These contracts execute automatically when
predefined conditions are met, eliminating intermediaries, reducing costs, and
accelerating the transaction process.
Cross-border Payments: Blockchain's
borderless nature and cryptocurrency integration simplify cross-border
payments, reducing fees and transaction times while eliminating the need for
However there are some very complex challenges that Industry
need to be aware and prepare itself before it can leverage Blockchain on
practical grounds -
● Regulatory Frameworks: The complex and evolving
regulatory landscape across different countries can pose challenges. Blockchain
applications often intersect with existing financial and trade regulations. To
harness blockchain's benefits, businesses must work with governments and
regulatory bodies to establish clear guidelines, ensure compliance, and
navigate potential legal hurdles.
● Interoperability: Blockchain solutions need to
interoperate seamlessly across various platforms and networks to achieve their
full potential. In the cross-border context, different countries and
organisations might use different blockchain protocols or platforms. Ensuring
interoperability is essential to enable frictionless data exchange and
transaction flow across borders.
● Scalability and Performance: As more transactions are
added to a blockchain network, scalability becomes a concern. The technology
needs to handle a growing number of participants and transactions without
compromising performance or increasing costs. Solutions like consensus
mechanisms, sharding, and layer-2 solutions must be explored to maintain efficiency
as the network expands
2. Energy efficiency is vital for
IoT devices with limited power sources. What are some innovative approaches to
optimising power consumption in IoT devices, and how do these strategies impact
the overall system design?
Answer from Vinod Singh:
efficiency is crucial for IoT devices with constrained power sources. Here are
some innovative approaches I have used or advised companies in IoT vertical to
try over last few years:
- Low-Power Hardware: Designing IoT devices with
energy-efficient hardware components, such as low-power processors and
sensors, is foundational. These components consume minimal energy during
operation and can significantly extend the device's battery life.
- Sleep Modes and Wake-Up Strategies:
Implementing sleep modes and wake-up strategies allows IoT devices to
conserve power during periods of inactivity. Devices can enter low-power
states and wake up only when necessary to perform specific tasks, reducing
overall energy consumption.
- Energy Harvesting: Integrating
energy harvesting technologies, such as solar panels or kinetic energy
harvesters, enables IoT devices to generate power from their surroundings.
This approach can supplement or even replace battery power, making devices
more sustainable and reducing the need for frequent battery replacements.
- Data Compression and Edge Computing:
Minimising the amount of data transmitted and processed by IoT devices
reduces their energy consumption. Data compression techniques and edge
computing allow devices to analyse and filter data locally, transmitting
only essential information to central systems.
- Adaptive Communication Protocols:
Using adaptive communication protocols, IoT devices can adjust their
transmission power based on factors like distance and signal quality. This
ensures that devices use the least amount of energy required to maintain
- Machine Learning for Power
Optimization: Machine learning algorithms can analyse usage patterns
and optimise power consumption in real-time. By learning from device
behaviour and user interactions, these algorithms can fine-tune power
management strategies for maximum efficiency.
these strategies create challenges in the hardware design and the cost of these
devices so not every strategy is suitable in every use-case.
Here are some
key consideration while designing hardware for these devices -
● Hardware Architecture: The choice of
energy-efficient components and hardware design considerations shape the
device's power consumption profile from the ground up.
and Software: Sleep modes, wake-up triggers, and communication protocols are
integrated into the firmware and software to manage power states effectively.
Experience: Optimised power consumption enhances user experience by extending
battery life and reducing the need for frequent charging or maintenance.
and Cost: Energy-efficient IoT devices require less frequent battery
replacements or recharging, reducing maintenance efforts and costs.
● Scalability: Power-efficient design is essential
for scaling IoT deployments without exponentially increasing power demands,
making it easier to manage larger networks.
3. Customer experience is a key
focus in insurtech. How can AI-powered chatbots and virtual assistants enhance
interactions between policyholders and insurance companies, streamlining tasks
like claims submissions and policy inquiries?
Answer from Vinod Singh:
Certainly, AI-powered chatbots have been in use for years,
but their effectiveness has often fallen short of expectations. However, with
the recent emergence of Large Language Models (LLMs) like ChatGPT and Google
Bard, the landscape is rapidly evolving, and AI-powered chatbots and virtual
assistants are poised to deliver more human-like responses than ever before.
This advancement is an exciting prospect for the future of customer experience
in the insurtech industry.
Traditionally, chatbots struggled to understand the nuances
of human language and context, leading to frustrating interactions and limited
capabilities. However, LLMs have changed the game by demonstrating an
impressive ability to comprehend and generate text in a manner that closely
mimics human communication. These models can engage in natural, context-rich
conversations, making interactions with AI-powered virtual assistants feel
advancements in LLMs have a significant impact on customer interactions in
insurtech. With these advanced models, chatbots can:
● Understand Complex Queries: LLMs can comprehend
and respond to intricate queries, enabling users to ask questions in a more
Contextual Responses: These models can maintain context throughout a
conversation, offering more relevant and coherent responses, which is crucial
for insurance inquiries and claims discussions.
Personalized Assistance: LLM-powered chatbots can analyze a user's history and
preferences to provide personalized recommendations and solutions.
Ambiguity: These models can better handle ambiguous or unclear queries, seeking
clarification before providing an answer, which is essential in insurance
discussions that often involve technical terms.
Self-Service: The improved capabilities of LLMs empower customers to perform
self-service tasks more effectively, reducing the need for human intervention
in routine inquiries.
Problem Solving: LLM-powered chatbots can assist in complex problem-solving
scenarios, guiding customers through intricate insurance scenarios and policy
● Enable Natural Conversations: Conversations with
LLM-powered virtual assistants feel more natural and human-like, fostering a
sense of ease and familiarity for policyholders.
4. Drawing from your experience,
how have insurtech companies effectively applied machine learning algorithms
and data analytics to develop more accurate pricing models for usage-based
Absolutely! Insurtech companies
have effectively applied machine learning algorithms and data analytics to
develop more accurate pricing models for usage-based insurance products in a
number of ways, including:
and analyzing large amounts of data. Usage-based insurance (UBI) products
collect data on a variety of factors, such as driving behavior, mileage, and
location, which can be used to create more accurate risk profiles for drivers.
Machine learning algorithms can be used to analyze this data and identify
patterns and trends that can be used to predict future claims costs.
personalized pricing models. By using machine learning algorithms to analyze
individual driving behavior, insurtech companies can develop personalized
pricing models that more accurately reflect the risk of each driver. This can
lead to lower premiums for safe drivers and higher premiums for riskier
real-time pricing updates. Machine learning algorithms can be used to
continuously update pricing models as new data becomes available. This allows
insurtech companies to provide real-time pricing updates that reflect changes
in driving behavior or other factors that may affect risk.
safer driving habits. By providing personalized pricing based on driving
behavior, insurtech companies can encourage safer driving habits. This can lead
to lower claims costs for everyone, which can ultimately benefit both insurers
Here are some specific examples of how insurtech
companies have used machine learning algorithms and data analytics to develop
more accurate pricing models for usage-based insurance products:
uses big data and telematics devices along with AI to power insurance companies
in UK to offer pay as per use or lower premium model for young drivers based on
their driving behaviors
Insurance uses machine learning to analyze data from telematics devices
installed in vehicles to create personalized pricing models for its UBI
products. This has allowed Root to offer lower premiums to safe drivers and
higher premiums to riskier drivers. As a result, Root has been able to grow its
customer base rapidly and become one of the leading UBI insurers in the United
also uses machine learning to analyze telematics data to create personalized
pricing models for its UBI products. Metromile's pricing model is based on the
number of miles driven, rather than the driver's age, gender, or credit score.
This has made Metromile's products more affordable for young drivers and those
with poor credit.
uses machine learning to analyze a variety of factors, including driving
behavior, location, and weather, to create personalized pricing models for its
home insurance products. Lemonade's pricing model is designed to be more fair
and transparent than traditional home insurance pricing models.
5. Fraud detection is crucial in
the insurance industry. How can advanced data analytics, anomaly detection
algorithms, and AI-driven models be utilized to identify suspicious claims
patterns and prevent fraud?
Absolutely, fraud detection is a critical aspect of the
insurance industry, and advanced data analytics, anomaly detection algorithms,
and AI-driven models play a pivotal role in identifying suspicious claims
patterns and preventing fraudulent activities. Here's some of the use cases I
have used in past and working on few of these at the moment:
● Data Integration: Integrating various data
sources, including policyholder information, historical claims data, external
data feeds, and social media insights, creates a comprehensive view of
individuals and their behavior.
Recognition: By analyzing large volumes of data, advanced analytics can detect
patterns that might indicate fraudulent activities, such as unusually high
claims frequencies or unexpected correlations.
● Predictive Modeling: Utilizing historical data
and machine learning techniques, predictive models can identify patterns
associated with previous fraudulent cases. These models can then predict the
likelihood of new claims being fraudulent.
● Establishing Baselines: Anomaly detection
algorithms establish baselines for normal behavior and usage patterns.
Deviations from these baselines trigger alerts for further investigation.
● Unsupervised Learning: Anomaly detection often
employs unsupervised learning techniques to identify irregularities that might
not be explicitly defined in the training data.
● Fraud Score Calculation: AI-driven models assign
fraud scores to claims based on various features, such as claim details,
policyholder history, and external data. Claims with high fraud scores are
flagged for manual review.
● Ensemble Models: Combining multiple AI models,
like decision trees, neural networks, and support vector machines, can enhance
fraud detection accuracy by leveraging different detection strategies.
Analysis: We heavily used this technique at Concirrus
Profiling: By analyzing historical behaviors and transactions of policyholders,
AI models can create profiles and detect deviations from normal patterns.
Monitoring: AI-driven systems continuously monitor user behaviors, raising
alerts for any sudden changes or inconsistencies.
Network Analysis: AI algorithms can analyze connections between individuals and
entities to identify networks of potentially fraudulent activities. This helps
uncover organized fraud rings.
Analysis: Representing claims, policyholders, and other entities as a graph
enables the detection of complex relationships and anomalies that might
6. What role do augmented reality
(AR) and virtual reality (VR) play in enhancing the online shopping experience,
and what technical considerations should e-commerce businesses keep in mind
when integrating these technologies?
I possess extensive experience in e-commerce, ranging from
global industry leaders like Amazon to smaller Shopify-based stores. Throughout
my journey, I've ventured into integrating AR/VR technologies within our
prototypes, and the potential these technologies hold is truly immense.
However, widespread deployment may still be around five years away, as the most
significant impact is expected to unfold during that period.
Recent research underscores this trajectory, revealing that
merely 2% of the global population currently has access to AR/VR headsets in
some capacity. As a result, e-commerce companies might encounter challenges in
demonstrating immediate ROI from these technologies. Nonetheless, it's
important to recognize that AR/VR are on a trajectory of evolution that mirrors
what transpired with smartphones. As costs decrease and accessibility broadens,
these technologies are destined to become commonplace, inevitably reshaping the
Having said that, Augmented reality (AR) and virtual reality
(VR) are two emerging technologies that have the potential to revolutionize the
online shopping experience. AR allows users to overlay digital information onto
the real world, while VR creates a fully immersive digital environment. Both
technologies can be used to enhance the online shopping experience in a number
of ways, including:
visualization. AR and VR can be used to provide customers with a more realistic
view of products before they buy them. This can be especially helpful for
products that are difficult to visualize in a traditional 2D format, such as
furniture or clothing.
interaction. AR and VR can be used to allow customers to interact with products
in a more realistic way. This can be done by allowing customers to virtually
try on clothes or furniture, or by allowing them to walk around a virtual
and training. AR and VR can be used to provide customers with educational and
training materials about products. This can be helpful for products that are
complex or difficult to use, such as appliances or electronics.
engagement. AR and VR can be used to engage customers and create a more
immersive shopping experience. This can be done by creating interactive games
and experiences, or by providing customers with a virtual tour of a store.
There are also some technical considerations that e-commerce
businesses should keep in mind when integrating AR and VR into their online
shopping platforms. These include:
● Hardware requirements. AR and VR require specialized
hardware, such as smartphones, tablets, or VR headsets. Businesses need to make
sure that their customers have the necessary hardware before they can offer AR
or VR experiences.
● Bandwidth requirements. AR and VR can require a lot of
bandwidth, especially for high-quality experiences. Businesses need to make
sure that their websites have enough bandwidth to support AR and VR experiences
without slowing down the user experience.
● Development costs. AR and VR development can be more
expensive than traditional web development. Businesses need to factor in the
cost of development when deciding whether or not to integrate AR and VR into
their online shopping platforms.
● User experience. AR and VR experiences need to be
designed for the specific needs of the target audience. Businesses need to make
sure that the experiences are easy to use and understand, and that they provide
value to the customer.
As the AR/VR headset market continues to grow, it is likely
that the percentage of the population with access to this technology will also
increase. It is possible that by 2030, as much as 10% of the world's population
will have access to AR/VR hardware
7. Traditional credit scoring
models are being augmented with alternative data sources. Can you elaborate on
the technical aspects of incorporating unconventional data, such as social
media activity or transaction history, into creditworthiness assessments?
I have had the privilege of collaborating with visionary
minds in the banking industry on this very subject. I firmly believe that the
banking sector has a significant opportunity for improvement by redefining how
credit scores are computed. Currently, credit scoring heavily relies on factors
such as consistent income, residential address stability, credit history, and
financial behaviors like timely credit card bill payments or savings. However,
the World Bank reports that approximately 1.7 billion adults worldwide lack
access to even the most basic bank accounts. This leaves them without the
ability to save, receive payments, or access credit.
fundamental challenge in traditional credit scoring lies in the absence of a
permanent and stable living address. According to the World Bank, roughly 1.6
billion individuals globally lack a permanent address. Consequently, they often
fall within the categories of "unbanked" or "underbanked,"
as they lack access to conventional financial services.
statistics underscore the urgent need to introduce an innovative credit scoring
approach that leverages alternative data sources. Numerous data points can be
tapped to assess the creditworthiness of individuals devoid of a conventional
credit history. These include:
● Telecom Data: Insights into individuals' cell
phone usage, encompassing call frequency and data consumption.
Media Data: Information from individuals' social media activity, including
spending habits and financial aspirations.
Data: Data indicating residential and work locations, valuable for evaluating
Shopping Data: Purchasing behavior in online transactions, providing insights
into spending habits.
History: A record of past bill payments, spanning rent, utilities, and mobile
Details about individuals' earnings and income stability.
Information on savings and investments.
Data regarding debt levels and management.
History: Insights into tenure and stability of employment.
Educational attainment level.
● References: Input from individuals familiar with
the applicant, offering insights into their financial responsibility and
8. Can you discuss your
observations regarding the technical challenges and solutions in building
scalable, cloud-based platforms for processing and settling insurance claims in
real-time, incorporating features like document verification and digital signatures?
Building scalable, cloud-based platforms for real-time
insurance claims processing requires addressing challenges related to
scalability, data integration and Data Quality, security, and user experience.
Implementing technologies like microservices, APIs, blockchain, and digital
signatures, while incorporating machine learning for fraud detection, can
provide robust solutions. These innovations can revolutionize the claims
process, making it faster, more secure, and user-friendly, ultimately enhancing
customer satisfaction and operational efficiency for insurance providers.
Here is the list of the technical challenges in building
scalable, cloud-based platforms for processing and settling insurance claims in
real-time, incorporating features like document verification and digital
volume. Insurance companies collect a massive amount of data, including claims
data, policy data, and customer data. This data can be stored in a variety of
formats and systems, which can make it difficult to integrate and analyze.
silos. Data silos are a common problem in insurance companies. This is when
data is stored in different systems and cannot be easily shared. This can make
it difficult to get a complete picture of a customer or a claim.
quality. Even if insurance companies have access to all of their data, it may
not be of good quality. This can be due to a number of factors, such as human
error, outdated systems, and inconsistent data entry. For delivering precise claims assessments
through modern technology, a substantial foundation of data is imperative.
Equally critical is the exceptional quality of this data to prevent
inaccuracies. The process of preparing vast volumes of data to be compatible
with analytics and AI demands substantial time and significant investment,
often reaching millions of dollars, when executed without precision
● Scalability. Insurance claims can be unpredictable and
can vary in size and complexity. A scalable platform must be able to handle a
large volume of claims without impacting performance.
● Real-time processing. In order to provide a good
customer experience, claims must be processed in real-time. This requires a
platform that can quickly and efficiently process large amounts of data.
● Document verification. Many claims require the
verification of documents, such as medical records and police reports. This can
be a time-consuming and manual process. A scalable platform must be able to
automate document verification to improve efficiency.
● Digital signatures. Digital signatures are increasingly
being used to authenticate documents and verify identities. A scalable platform
must be able to support digital signatures to ensure the security and
authenticity of claims.