Cognitive artificial intelligence is expected to be a big business as research firm IDC forecast that firms will spend approximately $77.6 billion on the technology by 2022.

However, organizations have to overcome often overlooked hurdles like linking disparate systems, to reach AI’s true potential, said Gagan Kanwar, director of business development at MuleSoft, during a recent webinar.

“Siloed data and lack of connectivity restrict AI’s ability to create real business value and turns it into little more than an expensive brain in a jar.”

When it comes to conversational AI, which incorporates machine learning and natural language processing, companies need to think beyond just deploying the technology in single-channel point solutions, added Kevin Fredrick, a managing partner at OneReach.ai and who also participated in the webinar.

Conversational AI is one of the prime ingredients in the fourth generation of robot process automation, which automates judgment tasks and consumes unstructured data while providing predictive and prescriptive analytics.

This method speeds up processes by democratizing access to data, according to Fredrick. “We think that bots eventually will be able to create themselves. If that is too far out there for you, we can start with more common use cases.”

OneReach.ai has identified more than 100 use cases for cognitive AI within most organizations like onboarding and training new employees as well as entering sales leads into a customer relationship management platform.

Firms wanting to reach the lowest hanging fruit might want to start with a conference-room booking or desk-side wi-fi management applications, he recommended.

“These internal uses are a great place to start if you haven’t deployed a conversational application yet since you can build the skills and culture for using AI without having to expose your customers to an underdeveloped experience,” said Fredrick.

It’s a matter of “when,” and not “if,” organizations adopt conversational AI, he noted. “It’s the natural evolution of the user interface- from physical to graphical to eventually conversational. Machines will communicate with you in the way you want to communicate rather than the other way around.”