A new group of unicorns is attracting the attention of technology investors. Last week, Automation Anywhere (AA), a developer of back-office robots, announced it had secured a $250 million investment in a round led by Goldman Sachs Growth Equity and New Enterprise Associates.
Believed to be the largest ever first-time investment in a technology company, the valuation of the Silicon Valley-based business now stands at $1.8 billion, comfortably surpassing the $1 billion threshold required to join the unicorn stable.
The investment also heralds a feeding frenzy around AA’s line of business: robotic process automation, or RPA.
RPA promises to automate the basic, repetitive office tasks usually delegated to humans, like sending invoices or allocating incoming emails to the right departments. An RPA bot can manage the work of three or four office employees for a licensing fee of about $8,000 a year, according to Forrester Research analyst Craig Le ClairOpens a new window , potentially providing labour savings exceeding $100,000 a year.
The industry describes RPA as “taking the automation out of the worker†to free them up for more rewarding work. Enterprises can either cut jobs or redeploy staff to add value in other areas such as customer service.
AA’s sky-high valuation comes on the heels of similar unicorn valuations for its two main RPA rivals, the Romanian-founded UI PathOpens a new window and the UK’s Blue PrismOpens a new window , both recently valued at more than $1 billion. These massive valuations indicate investors’ confidence that RPA will be rapidly adopted by enterprises around the world. All three RPA companies report significant contracts from major businesses: AA clients include Unilever, Google and Cisco, and UI Path is working with BMW, Dentsu and Huawei.
Automating Inefficiency
But RPA is still an emerging technology and has yet to fully prove its worth. There are disadvantages that potential clients must bear in mind. One is that if a process is already inefficient, simply automating will only make it inefficient faster. Adopting RPA alone is not enough – it must be accompanied by an in-depth time-and-motion study of how systems can be made more effective and relevant.
RPA works by observing and analysing human work processes, breaking them down into easily-automated chunks and replicating them through algorithms. But this exposes another limitation of RPA: as businesses modernize and adapt, they change their processes over time and new ways of working emerge. Human staff will need to work out these updated processes to provide systems that the bots can observe and imitate.
RPA is not a one-off automation solution, but a costly, ongoing development.
A potentially bigger problem is that RPA is one-dimensional – it can replicate a straightforward task as long as it contains limited choice and selection. Many mundane tasks require some form of human input, such as spotting errors in forms, deciding how to act on a request or dealing with complex steps. RPA is not well-suited to this, but another parallel technology known as cognitive automation is rapidly being developed to expand RPA’s scope.
Dealing with Unstructured Data
Cognitive automation augments RPA by observing humans at work and incorporating various artificial intelligence skills such as optical character recognition to read and classify documents and natural language processing to understand texts. This enables the automation of more complex tasks such as replying to complaints.
AA has launched a product called IQ Bot, which it claims bridges the gap between RPA and cognitive platforms such as IBM Watson. While RPA can process structured data such as spreadsheets and databases, it struggles with unstructured data such as emails, images and voice messages. Cognitive automation can make sense of this unstructured data.
Automation has been around for decades in manufacturing and is taking off in white-collar professions, but RPA is a blunt and not particularly intelligent instrument that needs to hone its cognitive abilities. While enterprises are keen to try it out, it seems too early to place bets on RPA as the future of automation.
AI systems from IBM Watson, Google or Microsoft Azure may prove more transformational in the long run with their greater cognitive capacity, but for the next few years at least, RPA will offer a simple and easy path to automating mundane tasks.