In the past fortnight, I chatted with an internet 'bot' (short for web robot) possibly thrice: to renew my vehicle insurance, report a delivery error on a food-tech app, and a Twitter response from an airline. I couldn't be entirely sure if it was an automated bot—some had names like 'Kalyani'—or an actual person typing out the response!
This by itself is creditable for the fast-evolving field of chatbots, especially because the chat experience is still as annoying as speaking with a customer support executive. (Effective bots mimic the human experience.) But there is value in the bot function, for example, when I clarified the expiry date of my insurance policy. The response was slightly delayed, but polite after checking my details (policy number, etc). Five years ago, I depended on a phone call or email query. Chatbots are changing that experience for customers.
A chatbot is a computer program based on artificial intelligence, which conducts a conversation using auditory or textual methods. “You can develop a bot, which is a computer program that replicates specific tasks users do,” says Ravi Venkatesam, founder of OnTrac, which helps companies improve their operational efficiencies.
For an enterprise, bots draw from the experiences and history of human actions carried out in its information systems—data from phone calls, emails, scanned documents, Excel sheets, etc. That's why India, with its legacy in the BPO/BPM industry, is becoming a hotbed for bot ventures. A lot of the back-office systems were managed by offshore professionals sitting in India. The IT industry knows a lot about enterprises across sectors and geographies.
While online services deploy chatbots to speak with customers, enterprises across industries are using it to make their internal workflows faster and more efficient. This is the robotic process automation (RPA) industry.
Value or Opportunism?
According to tech analytics firm Tracxn, the enterprise bots segment took off in the US from 2010. By 2013, ventures like WorkFusion, i.am+ and Wizeline had begun operations and raised $300 million between them. A couple like Sobot and Emotibot have sprung up from China too in the past five years. India has the likes of Gupshup. In all, chatbots have attracted $745 million from investors.
Many of these ventures' prowess is dependant on the API economy. They draw from data residing in other information systems, which are connected by APIs (application programming interface). But the action has really picked up in RPA, where investors have plonked around $1.8 billion in 15 ventures.
The RPA companies help enterprises identify less-efficient tasks that can be automated using bots. Many of these ventures were founded in the pre-Lehmann crisis era (2005-08), which is important because they have traversed a massive curve in terms of time spent on RPA and enterprise workflows. This helps them build and develop computer programs to replicate human actions across a large number of enterprise situations and cases.
Two companies—Automation Anywhere in the US and UiPath in Romania—have raised nearly a billion just between them. The former has the likes of SoftBank Investment Advisers, New Enterprise Associates, and General Atlantic backing it, while UIPath has raised capital from Accel, Sequoia, and Google's CapitalG.
This piqued my interest: why does this arena for productivity need so much capital? What's the hurry? The simple answer I heard from experts was the windows of opportunity for large-scale RPA deployments are slim.
The current wave is getting built on a decade of human tasks and actions on computer systems. Winning now will be important before the next phase of unique human activities on more intelligent enterprise systems. Those set of workflows will take another three or five years to form, which will be followed by another RPA wave.
So while there is opportunism, who wins this leg will be determined how many large enterprises trust the RPA companies with their workflow information.
I met Ankur Kothari, co-founder of Automation Anywhere, when he was in Bengaluru last month to discuss how they monetise. And, why do enterprises need ventures like Automation Anywhere?
The company has been around for 15 years. Five years ago, the management felt the automation industry was focussing only on “interoperable” systems or those capable of talking with each other. That was powered by API, and represented less than 20 percent of information systems. “We wanted to solve for people who interact with systems in enterprises the way you and I use the phone,” Kothari said. “We want to solve that 80% problem, where bots can mimic human actions like moving data from one place in an enterprise to another based on rule-based algorithmic decisions.”
Basically, use bots to shrink the time of transferring data within an enterprise and its ecosystem. Automation Anywhere is scaling up its headcount: from roughly 400 people in 2013, it now has 1,400 employees. The company expects to touch 3,500 employees by this year-end. Automation Anywhere charges clients for the product licence fee and/or the recurring expense of bots developed using its product.
Matter Of Productivity
Why do enterprises or clients need this? For now, it makes their processes more productive and will lower the value of human tasks, as bots automate them. For example, the high-volume but a manual task of typing a piece of information, and copy-pasting it to a defined set of Excel files. “RPA comes into the picture for anything users are doing in a predictable way to interact with the computer,” says Venkatesam.
A long-term reason clients are buying RPA products is it requires data to be structured, but a large chunk of their data is unstructured. Over time, enterprises can make their systems intelligent using AI tools, but only if the data is structured. “AI will be very focussed on a single task. You don't have an AI today that can do complex things,” Venkatesam adds.
Put another way, AI-based tools are as smart or silly as the data it is built on. Last week, I attended a diversity conference organised by Interweave, called InclusionNow, where an Amazon project got cited for its unconscious bias in recruitment. News agency Reuters had reported this in October last year.
“The AI-based recruitment engine was biased towards picking men,” said a conference participant, who works with another technology multinational. “In any AI-based solution, the data that goes in determines the results you get out. Recognising the bias, Amazon shut it down,” she explained. Yup, bots do hold a mirror to organisational frailties.
Tech Trail is a column that delves on technology in the Indian realm. Kunal Talgeri is a freelance journalist in Bengaluru. The views in this column are those of the author.
Published Date: Mar 07, 2019 02:03 PM | Updated Date: Mar 29, 2019 09:03 AM IST