Author: Mr. Sahil Gupta - Director (Growth & Ecosystem) - Techforce.ai How deep-tech is shaking business automation industry; An insider view
Before we delve deep into the impact of deeptech on the automation industry, let’s first try to define Business Process Automation and it’s evolution from standard scripting based automation to a more evolved and cognitive form of automation using deep learning and Artificial Intelligence.
According to Wikipedia – “Business process automation is the technology-enabled automation of various processes to streamline a business for simplicity, achieve digital transformation, increase service quality, improve service delivery or contain costs.”
Automation has been driving business success ever since the dawn of the industrial revolution with the first types of robots that were seen on the assembly lines in factories. Today, we are witnessing a similar trend in the information industry with software-based automation finding applications in every aspect of enterprises including multiple business functions like Finance and Accounting, Procurement, Human Resources, Talent Management, IT and Customer Experience.
According to a survey of global executives published by Deloitte, 73 per cent of respondents said their organizations have embarked on a path to intelligent automation: a significant 58 per cent jump from the number reported in 2019. Pandemic seems to have played its part in accelerating this transformation as many enterprises realized the need to build business resiliency when their business continuity protocols were challenged to the core. We will explore this a bit more in depth later in this article when we discuss case studies on how many organizations started leveraging digital workforce (or E-Workforce) as we can call it to work alongside their Human Workforce.
Unlocking true business value with AI and deep tech
We have read in the previous section that how businesses started leveraging process automation to handle most of their repetitive and labor-intensive processes. But until 2018-2019 most of it was managed through Robotic Process Automation (RPA) with almost zero to negligible intelligence built in to handle any form of unstructured data. As a result, businesses were not able to scale their automation to multiple processes or avenues which would yield much higher ROI.
In fact, while the RPA industry was buzzing with investment activities and emergence of global leaders raising billions of dollars from renowned VCs, many analysts started touting the same period as the death of RPA and many in fact labelled it as a fancy macro. RPA could not be applied to processes which were complicated and required multiple decision-making points that could lead to exceptions. Exceptions usually meant “breaking bots” which in fact made RPA quite unpopular in some circles. Even early adopters were not immune to these problems and many of them invested huge resources to scale and get success with their RPA programs.
RPA for sure needed an infusion of AI and deep tech to lead businesses to the “Promised ROI land”.
As we can see in the diagram below, this was not far off too as all the vendors realized the shortcomings with pure play RPA and started introducing AI into their products.
image credit : pwc
Some of the very initial forays and investments were made in the document processing space with leaders emerging and in some cases, even surpassing the massive valuations commanded by erstwhile RPA players.
While market leaders like UIPath, Automation Anywhere and Blueprism were busy expanding lead, the need for intelligent automation also saw multiple players emerge and challenge the incumbents on their own turf. There were specialists like ABBYY, Hyperscience among others in the document processing space and eventually vendors like Techforce.ai that emerged with a fully integrated automation stack comprising of AI capabilities like document processing, natural language processing, workflow orchestration and no-code development.
The period between 2019 – 2021 saw various partnerships forged between RPA leaders and specialist players to offer an integrated stack to customers while some RPA leaders continued to invest in native AI capabilities as well. Many vendors leveraged mergers and acquisitions as a channel to quickly build up their AI and deep tech capabilities, especially in process mining to offer their customers a way to identify which processes to be automated instead of relying on manual or instinct driven ways for automation. Some of the notable names to be mentioned in this list include Microsoft, SAP and ServiceNow and Appian. This indicates a clear market trend by notable product vendors to offer intelligent automation as a built-in feature; something that customers will growingly expect their vendors to offer natively instead of relying on 3rd party integrations through after sales services.
Human Workforce + E-Workforce = Autonomous Workforce
Early adoptions of process automation were mostly into the Finance and Accounting, HR, procurement (largely termed as back-office of the Enterprise). Even then, most of the processes usually dealt with data entry between multiple systems, report generation, data reconciliation among others due to obvious reasons that RPA as a standalone technology was limited to structured data and rule-based processing automation.
Even in the front-office (processes dealing with customer engagement), we saw basic chatbots that would do structured query processing and retrieving answers from systems using standard API calls.
Businesses must look at ways to break these silos between the front office and back office and empower their people across various business functions to collaborate better. HFS Research has termed this new state as Digital OneOfficeTM where an organization has fully embraced digital business models, enabled by an intelligent support capability, where barriers between front and back offices no longer exist.
Let’s explore what an AI and deeptech powered Intelligent Automation platform looks like. At Techforce.ai, we took an integrated view of the various deeptech capabilities that exist in the market and came out with a concept of E-Workforce. “E-workforce is digital twin of an organization’s human workforce and makes process operations resilient while performing manual work at scale.”
image credit: pwc
To work closely with your human teams and assist them with their day-to-day processes, the technology should have similar cognitive skills as humans. While presently AI may not be a match for human skills when it comes to vision or speech understanding, the fact that it can still support us by taking over a lot of our repetitive and mundane activities means we can have more time to focus on value added activities.
In its recent report McKinsey pointed out that about 60 percent of all occupations have at least 30 percent of constituent activities that could be automated. I am confident this number is slated to go up as cognitive capabilities of platforms improve over the time. This would mean humans would be working together with AI to get work done with AI and automation taking care of low to mid-level complexity tasks and decision making and only needing human intervention for most complex scenarios. This is what we describe as autonomous workforce where humans are relieved from the drudgery of repetitive, non-value-added activities and can focus on improving the processes and delivering better customer value at every step of the process.
Leveraging Conversational AI and Automation to transform Customer Experience
There are numerous examples of how businesses have leveraged conversational AI to solve their customer engagement challenges but erstwhile chatbots were very limited in functionality and could only deliver results that were available through standard API calls to backend systems. This means that customer would only be able to get about 2-3% of their requirements fulfilled digitally and would have to rely on the contact center agents for other queries.
Pairing conversational AI (both voice and chat) with RPA in the backend, enterprises are now dealing with this challenge and coming up with an ever-increasing bouquet of services that customers can digitally access now. While modern age startups like Zerodha, Groww would have digital first systems already, the traditional, larger banks with legacy systems needed a way to compete and deliver services digitally. Same holds true for other sectors where traditionally larger enterprises are busy thwarting competition from new age players and looking to digitally engage with customers.
Imagine getting your credit card blocked just via a chatbot on Whatsapp and zero human interaction. Getting your transaction statement delivered to you on-demand just by chatting up with a bot on the bank’s website or online banking channels.
We will discuss 2 different scenarios below on how different enterprises leveraged AI to deliver exceptional customer service.
Scenario 1 – Automating customer support via conversational AI and RPA
A financial services company was facing issues of ever-increasing inbound calls per month. Their main concern was to deliver better customer experience and improve the quality of interactions. The teams/human agents were working on peak capacity and struggling to keep up with the surge in calls. On an average, the contact center was receiving 30,000 inbound calls and emails in a month. The productivity was severely affected as the average handling time (AHT) stood at 30 minutes while the average turnaround time (ATT) took 24 hours. With a below average customer experience, the Net Promoter Score (NPS) was recorded at 6/10. This meant most of the customers were not particularly thrilled by the services, possibly resulting in high customer attrition.
The company decided to roll out WhatsApp based support for their customers and went live with about 30 of the top queries in the first phase. In the backend WhatsApp was integrated with an RPA engine to ensure even legacy systems were integrated and customer queries were addressed end to end. With automation, the team could focus more on issues requiring extra attention while the e-worker could manage routine queries expertly across chat and email. This allowed the human agents to build a healthy communication with customers improving the quality of live interaction. In just one month, the company’s contact center witnessed significant improvement with about 89% tickets automatically handled by e-worker, AHT reduced to 2 minutes, ATT to 4 minutes and above all the NPS rose to 8.5/10.
Scenario 2 – Reducing IT Support costs with conversational AI and automation
The customer’s IT helpdesk received a large number of tickets each month on L1 support and FAQs. Amid remote work and the ever-increasing volume of tickets, the turnaround time increased and there was no time to focus on core IT issues. IT leadership wanted to reduce L1 and L2 tickets handled manually by automating the responses to high volume low cognitive requests.
After deploying a conversational chatbot over Microsoft Teams in less than 4 weeks, the organization enabled over 8000 employees to access the AI helpdesk and seek help on various L1 and L2 issues. Additionally, offered L2 automation leveraging RPA & APIs to quickly redress common issues like SAP updates, account unlocks, s/w installation and password reset.
Deeptech based Computer Vision for intelligent document processing
We discussed earlier how in its initial phase, automation was only limited to rule-based and structured data processing with very limited or zero ability to process business documents. It had very limited applications in the business and hence ROI did not match up to the investments most of the times.
Introducing intelligent document processing (or computer vision powered by deeptech like ML and AI) has opened numerous applications across various processes involving documents like vendor invoices, sales orders, insurance claims, customer KYC documents, financial reports, travel bills, contracts and agreements. As a result, every industry whether Financial Services, Manufacturing, Healthcare, Insurance, Legal, Retail, Hospitality, Wealth management, e-commerce and multiple others have started deploying intelligent automation.
Vendors are deploying multiple technologies like AWS Textract, Azure ML, Google Vision API on cloud as well as multiple open-source tools like Tesseract, Open CV, others to keep improving the accuracy of their models. Training on datasets with millions of samples is bound to increase accuracy for these models. However, handwritten documents still remain a challenge in this space but even with the currently available solutions, businesses have unlocked greater productivity. Legacy Financial services players are seeing their documents from over hundreds of years old being digitized, classified and archived while industries like manufacturing, retail are using document processing for invoice and sales order processing to speed up vendor payments.
Let us look at a case study on how a large Fortune 500 manufacturer transformed their accounts payable process for over a million invoices per year through intelligent automation:
Business Problem: This Fortune 500 company was struggling with a lack of standardization across processes and applications which resulted in silos, hampering the overall operational efficiency. Further, the users faced challenges in managing exceptions and there were poor collaboration across business stakeholders due to the vast scale of operations across geographies. Furthermore, the invoices were paper-based which required a lot of physical movement across departments. This resulted in low control over the invoices, delayed payments and higher penalties. Therefore, the organization realized the need to digitize invoices and enable paperless processing by minimizing human intervention and eliminating error-prone tasks.
Solution: The invoices received at any location are captured and their scanned images are sent to a common mail room. Relevant data and information from invoices are extracted using deeptech based computer vision models, indexed, and validated automatically with a provision for human approval. Intelligent automation allows for automatic matching of invoices against purchase orders and goods receipts. The new system has empowered the previously over-burdened team to focus on addressing pressing issues by freeing up their time from time-consuming work of data entry and processing invoices. This has also resulted in superior quality of work with 100% accuracy and business rule execution leading to improved supplier satisfaction due to timely payments.
Benefits of Intelligent Automation
Let us look at some of the reasons why AI powered process automation is best for your business:
- Significant reduction in operational costs
- Dramatic reduction of processing and service delivery time through automation
- Eliminating of human errors and improved consistency
- Flexibility and scalability to manage fluctuating workloads
- Increased customer satisfaction and retention
- Better Business Continuity and resiliency to deal with disruptions
- Fast and reliable service delivery
- Rapid individualized contact responses
- Choice of communication channels
- 24/7 service availability