Robotic process automation (RPA) is an umbrella term for a group of technologies that enable line-of-business users to set up, launch and administer virtual workers without relying on an IT department. RPA tools can link different applications together and automate routine tasks so they don't require direct human intervention. The general methodology for an RPA deployment is two-fold: There's a user interface allowing end users to craft automated workflows and a suite of scripts that automate routine and non-varying data input or data processing work. RPA platforms are increasingly being augmented with machine learning and with the idea that the "robots" will be able to iterate routines and optimize how they complete their data processes.
RPA is a small but fast-growing software market. Gartner estimated the segment’s worldwide revenue at approximately $850 million in 2018, representing a 63% year-over-year growth rate. RPA is appealing from a cost perspective because it frees up workers to do more complex and less routine work, reduces overall labor costs, and frees up IT professionals so they're spending less time implementing and supporting middleware for their colleagues. It also appeals to business users who see it as a way to bypass the usual IT processes of acquiring and integrating new technologies into an enterprise's current legacy stack — RPA platforms do all the process automation and data integration for them.
However, RPA has a few drawbacks. The ease with which it can be implemented to automate work depends in no small part on whether the organization's data is easily assimilated into any automation models.
Last year, Ted Friedman, distinguished VP analyst for Gartner, said, “As organizations accelerate their digital business efforts, poor data quality is a major contributor to a crisis in information trust and business value, negatively impacting financial performance.” It also impacts overall operations and machine learning. According to Alegion, a training data provider for AI and ML initiatives, a majority of enterprise organizations trying to use machine learning at work have run into problems with data quality, the data labeling required to train AI and building model confidence.
If an organization has a vast trove of unstructured data, such as handwritten data, images or audio content, it requires more complicated and expensive AI-based technology to automate the process efficiently.
The other big drawback: Most of the low-hanging automation fruit — structured and repetitive tasks — has already been automated or is being automated through scripting, macros or simple RPA deployments. Consequently, RPA could include significant planning and implementation costs, because what's left for an enterprise to augment via RPA is much more complex than earlier automation tasks.
Our list of the top 10 RPA platforms has been selected using criteria to winnow down entrants in the RPA space to those that best met the following criteria:
Does the vendor offer a programmable bot solution? Being able to customize and refine robots based on an individual customer's business processes was a must-have for our list.
Is the vendor enhancing its automation tools and processes with machine learning or other cognitive tools? Having bots that can correct themselves and adapt their routines based on learning from the data is critical. It's also future-proofing any RPA platform offering; natural language processing and image recognition are going to be must-haves soon, if they’re not already.
Does the vendor have a wide range of experiences across a wide range of industries? Vendors that have successfully figured out how to make their services and tools work in different industries will have a better likelihood of being able to make an RPA implementation work. They'll also be more likely to have internal processes in place to figure out how to assess a customer from a new industry vertical and see where RPA works with that industry's needs.
Does the vendor have an active support plan for customers and an active user/vendor community? Although the appeal of RPA is the set-em-and-forget-em model of automating work and then lightly monitoring the robots, there is something to be said for being able to turn to fellow customers or vendors in the same space to see what's possible in deploying and optimizing RPA platforms as offered via a specific vendor.
Are customers satisfied with the deployment and support the vendor provides? A comparatively smooth implementation means less business downtime for the customer and faster implementation of RPA benefits for the user.
Is there a central dashboard allowing customers to monitor the workflow metrics and thus assess operational performance? There's no point in deploying RPA to optimize the enterprise if there's no transparency into how the deployment is performing.
Without further ado, here’s the list, presented alphabetically (requires registration).