Smart Judiciary: Justice with AI

An alarming number of pending cases has plagued India for a very long time. While the judiciary and the legislature have taken significant steps, such as Alternate Dispute Resolution (Mediation and Arbitration), People’s Courts, Fast Track Courts, etc., a lot more work is needed to reduce the backlog of ~4 Crore cases. India can benefit from AI technologies being used in different stages of the judicial process to create a streamlined and speedy resolution process.

India is the world’s largest democracy with a population of over 136 Crores. It comes as no surprise that we often face a lack of resources in every sector, including our Judiciary system. A judiciary process goes through several stages wherein police, courts, and corrections operate in an interconnected form to lodge complaints, investigate cases, initiate judicial proceedings at courts, collect statements from the accused and defendants, analyze case facts, and order judgment. This labor-intensive system, which is largely non-digitized, has made the processes hard to administer and even harder to navigate for people. It is therefore essential to utilize technology to tackle this issue effectively.

The role of AI in delivering speedy justice

Artificial Intelligence (AI) has transformed various fields and functions, such as healthcare, banking, automobiles, etc., and is gradually shifting how different organizations operate. Data and vision analytics, machine learning, predictive systems, and automation are emerging AI technologies that have contributed to this evolution. These technologies can help ensure sustainable justice delivery and reduce the backlog of pending cases through streamlining processes. These could be routine and repetitive processes (such as documentation) or complex ones (such as witness cross-examination).

Judiciary in several developed countries such as the USA and Canada already utilize AI to help the judge evaluate bail and parole appeals. Such measures will reduce the overall time a particular case takes for completion and help lower the corruption surrounding the paperwork and reduce the margin for human error, thus alleviating the pain an average citizen goes through in the process.

The judicial process in India has already taken some steps in this direction. One such instance is marking the presence of under-trial prisoners through video-conferencing. This is convenient for the person who no longer needs to travel long distances for deposition while also saving public money in the process. Digitization of court records and updating court orders on the website is also a way for the court to use technology to streamline the process.

Five areas of judiciary where AI can create an immediate impact

While the judicial system has started navigating in this direction, there are many miscellaneous tasks performed by different entities of the system, where AI can perform or assist in some of the fundamental functionalities. These include:

To summarize, AI-powered machines can be used in various stages of a trial by helping the police, the lawyers, the judges, and in turn, the citizens who are the most affected by the delays of the system. When this is done, the noble cause of ensuring effective and sustainable justice to the masses shall be achieved.

References

The Contact Center of the Future: Driven by RPA, Transformed with AI

One of the most used buzzwords in 2020 was undoubtedly the “new normal.” The pandemic’s initial business impact was indeed unwelcome for most enterprises, but the consequent cost pressures and fluctuating demand have brought about a shift in their focus. Global businesses are now turning to digital channels at every touchpoint to deliver an unparalleled customer experience. At the heart of these customer touchpoints lies the customer service (CSM) function or contact center, as it is popularly known.

“All our customer representatives are busy; please hold the line.” Sounds familiar?

Across sectors, whether it’s transportation (Airlines, cab services), BFSI (Health and Life Insurance claims), or Retail (eCommerce goods and groceries), huge call volumes put tremendous pressure on the CSM workforce. This results in long average wait times (AWT’s), unacceptable customer experiences, and subsequently, high customer defection.

However, due to the pandemic, customers are also shifting to online channels. Recent months have seen an uptick in consumer spends via online platforms. In the United States, the e-commerce industry accounted for 33% of the total retail sales by July 2020, much higher than a forecasted 24% in 2024. In Europe, overall digital adoption increased from 81% to 95% during the pandemic . Online customer interactions have reached levels never seen before; the real question is, are most enterprises adapting fast enough?

Hyperautomation is the “new normal”

RPA has generated double-digit growth consistently in software revenue over the years. AI-augmented automation or Hyperautomation is not a new term; it has been used heavily in the context of process mining and discovery. While process discovery will continue to be a highly sought-after capability, the focus is shifting gradually towards augmenting traditional, deterministic automation with AI – especially in document processing, image recognition, voice and text analysis, and machine learning algorithms.

AI-augmented automation focuses on increasing the scope of automation by leveraging AI to make more sense out of unstructured data-heavy on text such as emails, customer feedback surveys, reviews, and social media posts. Since unstructured content forms the staple of most contact center backend systems, it is easy to infer why this solution is touted as a veritable goldmine of untapped insights.

AI use cases for the contact center: Low touch, more personalization

Banking on document processing

Processes that are traditionally paper-based require significant manual intervention to interpret each document and enter data into a backend system. Banking processes like e-KYC and mortgage processing are document-intensive and hence time-consuming and error-prone. Automated extraction of key information from documents allows service desk agents to close applications fast and revert in case of additional document requirements, thus speeding up the end-to-end process.

Handling insurance claims using image classification

Service desks at top insurance firms have two key challenges: to reduce handling time to process claims and to pinpoint fraudulent claims from valid ones. Damage assessment for auto and real estate claims is a manual process that results in significant inspection costs. Image classification algorithms can recognize dents, scratches from customer images and provide a damage assessment report in minutes, which can then be reviewed for exceptions by a human-in-the-loop agent. This brings the lead time to process a claim from days to a matter of minutes and reduces the cost-per-claim for insurance firms.

Virtual agents take centerstage

Voice and chat capabilities have been an integral part of the contact center ecosystem for many years now. But the capabilities of these components are primarily limited to handing over data from the customer to a contact center agent using text translation. Increasingly, these bots are being augmented using natural language processing (NLP) and natural language generation to interpret incoming text and respond, as a human would. This reduces each service agent’s burden and limits human involvement to the most critical and urgent issues.

Sense from sentiment

In industries with lower switching costs and multiple alternatives, such as airlines and eCommerce, creating a personalized customer experience becomes paramount to retain both mindshare and market share. Identifying the essence of a customer complaint or feedback using sentiment analysis models can help classify content into appropriate buckets and rout them to appropriate teams for faster redressal. This lets AI and automation take care of the essentials while the contact center agent can focus on a more personalized and empathetic customer conversation.

Customer service agents also use self-learning AI to get insights on cross-selling and upselling opportunities using machine learning recommendation algorithms. These algorithms analyze historical purchase patterns and spend patterns to suggest products that the customer is more likely to buy.

The race to the finish line

It is becoming increasingly clear that enterprises are looking at the contact center as a revenue generator. Solution providers are investing in big partner ecosystems to deliver on the AI promise via customer-centric use cases. The pandemic has ensured that siloed operations of business functions like customer service, sales and marketing, and operations are no longer feasible. With remote operating models and a digital-friendly customer, enterprises will need to rapidly test and deploy AI-augmented automation solutions to deliver pre-pandemic levels of customer satisfaction. If 2020 was about business resilience, 2021 will be about using all that self-learning to chart a road to recovery and create new opportunities. Isn’t that ironic?

References

No Room for Complacency when it Comes to Compliance

DNL Trust Bank has been driving a successful Robotic Process Automation (RPA) program with their Automation Centre of Excellence (CoE) at the helm. After one of their quarterly reviews, CFO Timothy Davis was pleasantly surprised looking at the excellent numbers around efficiency gain, savings, and time to value generated through the RPA program. She recalled the days when the bank started its RPA journey. Their CTO Mark Higgins did his due diligence to get it right the first time. From setting up the Automation CoE to scientifically prioritizing the suitable candidates for automation1, he took the right steps and was now reaping the benefits of a well-planned automation journey.

Besides the superlative metrics, what intrigued Timothy more was the adherence to compliance. She had been worried in the initial days. How would the journey take shape when bots carry out the processes instead of humans? How would the existing, stringent controls and processes manifest once the bots take over?

Mark had jokingly quoted science fiction author Isaac Asimov’s First Law of Robotics to Timothy, “A robot may not injure a human being or, through inaction, allow a human being to come to harm! Then why do you worry, Timothy?”

But adherence to compliance and statutory guidelines is no joke, especially when it comes to financial institutions. Mark understood this, and hence as a part of his due diligence, early emphasis was on identifying the products’ capabilities to adhere to such requirements.

Let us dive a little deeper to understand what these controls are and how a good RPA product complements them.

Identity and access management

Segregation of Duties is a primary requirement considering some of the stringent controls like the Sarbanes-Oxley (SoX) Act. Segregation of Duties is trivialized when an organization’s workforce has bots working alongside humans. With the introduction of concepts like Automation Singularity2, segregation of duties within the system becomes very important. Activities performed by bots and humans need to be clearly defined, tracked, and recorded for audit purposes. RPA platforms like AssistEdge track the end-to-end workflow for each transaction in a consolidated manner.

Detailed role-based access controls (RBAC) in the automation platform ensure controls for privileged accounts within the RPA platform. An in-built secure credential vault stores passwords in an encrypted format. The ability to integrate with external credential vaults like CyberArk allows the flexibility to select the desired vault for securing the passwords. As a governance best practice, bot credentials’ exposure must be to a limited user group.

Regulatory requirements

As per IPE requirements (Information Produced by an Entity) under SoX compliance, the organization must appoint a custodian for all standard out-of-box reports and custom reports to define the data across these reports and consolidate and share these with the auditors. RBAC capabilities of the automation platform allow you to designate the custodian(s) and assign relevant accesses.

In addition to this, it is important to allow the business users to decide what information should be captured in logs as per the guidelines and requirements. AssistEdge’s capability to customize what gets logged along with different levels of logging comes in handy at this point. It also gives the automation designer control to achieve data lineage and traceability easily.

Incident management

It is paramount to capture actions taken by the bot and actions performed to the bot. Any intentional or unintentional change made to the automation process configuration, schedule, or triggers has a domino effect. It affects SLA adherence, introduces processing errors, and eventually results in massive financial losses for the enterprise. AssistEdge provides consolidated logs of both actions by and actions to the bots within the platform.

It captures applications’ log in by the bot and step-by-step records of each transaction. Also, audit logs capture the edited data points and modified documents. The platform allows the download of end-to-end logs for each process and its associated transactions. Those charged with internal compliance (e.g., internal audit function, IT compliance) are sensitized about these factors. Protocols are put in place to maintain an updated listing of bots and establish a standard operating procedure for all updates to processes/ controls. The change management ensures that required guidelines reflect in bot design, where necessary. AssistEdge also allows the configuration of restricted access modes for Virtual Machines in bot farms. This feature counters unauthorized access and misuse of bot credentials in target applications.

Conclusion

These capabilities might appear rudimentary when looked at in isolation. But the absence of these in your enterprise’s automation program could be damaging. An organization needs to ask what’s at stake when they fail to meet even one of these compliance guidelines. Many RPA vendors, who do not have an enterprise mindset, tend to overlook these capabilities. It is the onus of the Automation CoE to stringently evaluate the automation tool selecting it for their enterprise. Along with scalability and reliability, security and compliance are the foremost pillars on which an enterprise can build a formidable automation program.

References:

Disclaimer: All characters and events depicted in this blog are indicative. Any similarity to actual events or persons, living or dead, is purely coincidental.

Supercharge Your Supply Chain – The Network Effect

“It’s said that something as small as the flutter of a butterfly’s wing can ultimately cause a typhoon halfway around the world.” – Chaos Theory

In complex, global supply chain ecosystems, disruptions for a partner on one side of the world can adversely impact their manufacturing partners across the globe. The pandemic brought this dependency to the forefront during 2020. Inventory stuck at closed borders and sales crippled by the closure of retail outlets brought entire value chains to a grinding halt1.

The pandemic was a clarion call to break down operating silos. The Supply Chain industry needed to reinvent its model, and the path forward was to build resilience. The ability to switch supply lines at the touch of a button would ensure the value chain preserved its integrity while taking corrective actions to resurrect downed partner operations.

How would an enterprise go about building and operating such an ecosystem, especially one that has relied on trust and performed like clockwork for decades?

The network effect

Demand signals, and the ability to accurately forecast and deliver, thrive on data constantly exchanged amongst the various tiers of a value chain. With the number of partners running into the thousands, such data interchange relies on links established between direct points of interaction. But what if these links extended to the entire ecosystem? Ambitious? Yes. But certainly doable.

What are the drivers to enable such a network, and what goes into building one that stands the test of adverse circumstances?

Knowing is half the battle

Amongst the wonderful things to come out of the 1980s, GI-JOE’s war cry flies high. Staying informed is the first crucial step in building or sustaining a thriving business.

Organizations can no longer claim disparate systems of records at partner sites as a black box. With interoperable systems and APIs at the forefront, even partners with low technological maturity can inform stakeholders of their supply and stock levels.

Forecasting needs accurate demand signals. Getting demand signals through to all stakeholders, direct and indirect, is key to ensure everyone has an equal stake at the table. A network is all about the interchange of such information. Data feeds from Bangladesh are transmitted to Birmingham-based factories, letting them know of the raw materials shipped. Simultaneously, the sales headquarters at Boston knows that the next marketing campaign is on track based on production queue status at the factory. Disruption in shipping channels at the Suez? The network fulfills an alternate supply – Bolivia steps in to cover the hold-up.

Insights are as good as the underlying data. And so, establishing a fail-safe mechanism for data interchange is a non-negotiable step 0 of the process.

Fixing gaps requires identifying them

Widespread partners can now be connected via the network, where there is seamless exchange of data, and prediction algorithms churn out insight after insight. With so many moving pieces, a chain is as strong as its weakest link. Suppliers’ inability to fulfill their orders from a geography puts stress on all other parts of the ecosystem—a domino effect.

Resilience is built on trust, and in a network, data drives this trust. Supplier scores are a mechanism to ensure the best of partners (or the closest alternate) is always available to protect the network’s sanctity. The scoring algorithms take into account multiple factors, primarily a partner’s track record, their tier-2 and 3 suppliers’ efficiencies, as well as macro-economic factors.

A common question that stakeholders pose when it comes to a network is: What’s in it for me?

Apart from the transparency and benefits from leveraging economies of scale, the notion of new markets opening up to a partner purely by virtue of a multi-enterprise network is unparalleled. Imagine breaking beyond the barrier of the 6 degrees of separation. With dynamic scoring algorithms, intelligent order routing engines, and predictive insights running atop the network, it levels the field for even small-scale partners while boosting top-line options for the larger ones.

Breaking the fourth wall

While several enterprise networks end at last-mile delivery, it is time to remember that ‘consumers’ are your partners too. Direct-to-consumer (DTC) models have taken off and, for several organizations, serve as the leading source of revenue. How, then, can these partners be unaccounted for?

What started as a means to directly reach consumers for their daily personal needs has evolved to even EV auto-makers going direct, shaking up the traditional fossil-fueled personal vehicle landscape. Consumers have leveled up to brand ambassadors even as online ad/promo spending budgets have shot through the roof.

Connecting with the end consumer has often remained the holy grail for principal brands. The network simply gives the right set of tools to enable the connection. By understanding consumer spends and market share projections, business applications built atop the network allow brands to fine-tune their strategy on the fly—all at the touch of a button. Bringing customer loyalty and membership as a use case to the network has resulted in better adoption and value realization.

Conclusion

Multi-enterprise Supply Chain Networks are not new. They’ve existed for quite a while now. However, most networks do not have much to offer, purely putting together building blocks – riddled with aging tech and still operating under silos, all under the guise of an interconnected system. The evolution of such networks is underway, and it brings with it a need to not only modernize the underlying infrastructure but to have point solutions built atop the network, which leverage the powerful data generated from near and far.

Pay-as-you-go and subscription models offering insights not only help the primary brands but even the smallest of partners by opening up new avenues of revenue and growth.

The future of the Supply Chain revolution has already begun. Come on aboard!

References:

AI & Automation: Success Stories in Manufacturing

The advent of Industry 4.0 brings with it the extensive use of smart digital technologies, like the Internet of things, cloud computing, robotics, software, automation, and cognitive technologies such as AI & machine learning, in manufacturing. Enterprises are becoming more open and aggressive in using these technologies to improve operational efficiency, reduce the mudai involved, and go lean to maintain their competitive edge. Manufacturers realize that digital technologies are essential for them to reduce costs, improve the bottom line, and also add to their topline growth. Resultant smart factories, better insights into processes, and optimized supply chains make manufacturing more efficient and poised to deliver better outcomes for industry players and customers alike.ii

In this article, we look at some use cases of manufacturers using the power of Automation, AI, and Applications of AI to become more efficient and customer-focused.

Unlocking process efficiency

A leading aircraft manufacturing and parts distribution company in the US, dealing with hundreds of suppliers and customers, struggled with its Purchase order (PO), Sales Orders (SO) processing. The existing process, handled by a large operations team and sales representatives, was predominantly manual and error-prone. Adding to the challenge was the complexity of the PO itself. A single aircraft contains lakhs of partsiii , and it’s normal for a single PO in the aviation parts business to have hundreds of line items to place an order for thousands of parts.

The process could be made more efficient with Intelligent Document Processing and Robotic Process Automation (RPA). EdgeVerve used Nia DocAI combined with AssistEdge RPA and automated the entire process.

As a result, the customer could eliminate the as-is 2-step process of data entry and validation, extract data with 80-90% accuracy, reduce AHT by 50%, and automate over 16,000 transactions per month. This saved over 43,000 hours of manual effort and subsequent costs.

Similarly, one of the world’s largest electronics companies and a leader in healthcare technology was looking to harness the power of digital transformation to accelerate its next phase of growth. They were looking at RPA to automate their Finance Planning & Accounting (FP&A) operations in the Record to Report (R2R), Order to Cash (O2C), and Procure to Pay (P2P) processes as well as Reporting and Consolidation activities. AssistEdge RPA automated 35 complex use cases saving 32,000 person-hours.

This implementation is one of the most complex, large-scale, and successful RPA implementations across the industry, with 400+ bots running in a High Availability setup 24X7 and more than 90% accuracy.

Driving new growth

When applied to client prospecting and building sales intelligence, the use of intelligent technologies could act as a new growth engine for the customer. A leading Aerospace Parts Manufacturing company relied on manual analysis of past trends to identify cross-sell opportunities. Their existing process had limited intelligence and relied only on internal data sources for building intelligence. This led to inaccurate opportunity identification.

The EdgeVerve team helped improve the process and make it more accurate. Using the NIA platform, we blended information from key internal and external sources, especially on projected customer fleet and its characteristics. It contextualized the ML-based recommendation system built by correlating the purchases from customers with similar fleet profile AND products strongly associated and are most likely purchased together. Products-to-aircraft associations were generated as additional intel for the sales team.

As a result, the customer could tap whitespace opportunities for 46 priority customers in the EMEA region. New verified opportunities worth USD 8 million were created annually for the EMEA region, and recommendations achieved an average hit rate of 70%.

Improving resource utilization

A leading Forklift Manufacturer in the US struggled with inefficient, manual processes. Precious SME and associate time, which could have been deployed towards business growth, was spent on doing mundane and repetitive tasks, leading to a lot of muda. The client deployed intelligent automation to optimize processes for its Fleet, Service Parts, and Dealer Development departments.

The Fleet department’s invoice coding process was one of the automation candidates. The process involved a rule-based validating and coding of invoices received from dealers. The process calculated labor hours and then either approved the invoice or sent it for manual review. This entire process, which involved reading from the repair description, was automated using AI technologies like NLP and text analytics. Machine learning algorithms were used to predict the labor hours for the repair. The process is now touchless and auto-approves invoices. Exceptions that require manual review also have a note to the reviewer that notifies them of the issue. This saves time and also has a saving potential of USD 155,000 per annum.

Leveraging AI and automation for your business

Technologies like AI and automation have the potential to transform your enterprise operations and manufacturing supply chain processes. The need is to follow a structured approach to identify these opportunities, prioritize the ones with higher ROI, and over time make your organization processes Agile and lean – in turn improving the customer experience and realizing the true potential of Industry 4.0.

References