Business process automation (BPA) is undergoing a major evolution in the generative AI era, thanks in part to AI startup Kognitos.
Kognitos today announced its new self-service approach to enabling organizations to use generative AI for business process automation. Kognitos has been building a platform that allows organizations to use natural human language to define and enable BPA. It’s an approach the company detailed at the VB Transform event last month. The new offering, Self-Service Generative AI for Centers of Excellence and Finance Organizations, extends the company’s platform and is the first time the AI startup has offered self-service, as Kognitos aims to make it even easier for business users to enable automation.
Over the last decade, business automation has been enabled in part by technology known as robotic process automation (RPA), but according to Kognitos RPA hasn’t been able to solve a broad spectrum of automation problems in large enterprises.
“What we are doing is bringing the power of generative AI to actually tackle that problem head-on and deliver the true promise of what RPA should have done for business processes,” Kognitos founder and CEO Binny Gill told VentureBeat in an exclusive interview.
Why self-serve generative AI matters
To date, Kognitos has helped its customers with their initial automation in a hands-on way. The basic idea with the new offering is to provide an easy on-ramp to automation that business users can run on their own.
Gill explained that some organizations have built what are known as Centers of Excellence to help get RPA implemented in years past. The goal with Kognitos’ new offering is to reach out to these internal Centers of Excellence and enable them to use gen AI to build and deploy business process automation capabilities.
With the self-service model, Gill explained, business users can for example use the Kognitos service to extract data coming out of an invoice or a purchase order and use it as part of a larger business process.
Why gen AI for process automation is more than a replay
Among the many ways business automation has been enabled in the past is with replays — recording a user’s screen to create a simple replay of a set of actions. Gill said that Kognitos will now allow organizations to use natural language to explain what they want to do.
He explained that using generative AI with natural language is a more scalable, resilient method than replaying a recorded operation. A replay can create brittle results, according to Gill, since the user interface and workflow of a business process can change over time.
Another challenge with a simple replay is that it can’t handle branching logic very well. Branching logic is when an interface has a set of conditions and options for the user, such that if a user presses one option, a particular set of options is offered, and if they select another, a different set is suggested.
Modern software is also commonly built with application programming interfaces (APIs), and in Gill’s view, there is no easy way of recording what the APIs need to do through clicking and recording those actions.
Human language is the key to business usage
A key challenge of RPA, according to Gill, is that it often requires specialized skills and some custom programming to fully enable.
Kognitos uses natural language with an engine known as the Human Language Interpreter, which is now being updated to version 2.0 alongside the self-service launch. Natural language processing (NLP) enables AI to understand natural language. Gill said that Kognitos’ Human Language Interpreter uses NLP at a foundational level, and then goes beyond typical NLP capabilities.
Understanding the business process context is key to the Kognitos Human Language Interpreter’s effectiveness. For example, a given statement could be interpreted in different ways. Gill said that the interpreter figures out the context of what the user is running and creates a knowledge graph of the world around it. Based on that knowledge it will better understand what the prompt is really about. If it isn’t able to clearly interpret the prompt, the system will ask the user a followup question to clarify.
Exception handling and dealing with issues when they occur in a process is another area where the Human Language Interpreter helps.
“We are trying to explain things to the layman so whenever something bad happens, we don’t want to [have to] pull in an IT guy,” Gill said.
Gill explained that Kognitos aims to make it easy for users to understand why an issue occurred, and provides prompts to help with remediation.
The new self-service platform is targeted at Centers of Excellence and finance functions, but the company has hopes for an even wider audience in the future.
“The goal is to bring this power to a billion business users. That’s our end goal and that’s our vision,” Gill said.