<img src="https://ws.zoominfo.com/pixel/Mpn8QOHA6F04ONzxJD62" width="1" height="1" style="display: none;">

5 Questions About Pairing RPA & AI for Intelligent Automation

Artificial Intelligence is one of the most talked-about technologies in the 4th industrial revolution. However, we often find ourselves looking at misleading information and 'AI-washed' content that can generate confusion to non-technical audiences.

To cut through the noise, we asked Danilo McGarry, advisor to the EU commission of AI Alliance and long time Automation thought leader, to join UiPath's Chief Evangelist, Chet Chambers, and JOLT's SVP of Ops, Brett Fraser, to present the 'How to Marry RPA and Artificial Intelligence to Drive World Class Automation' webinar, a presentation that centered on what exactly is AI, how is it different from RPA and how do the two technology categories complement each other to enable greater business outcomes. 

This webinar was part of our Hyperautomation Webinar Series, you can sign up to the series here and access the replay of this webinar here.

During the live Q&A session of the webinar, our attendees asked important questions regarding RPA and AI. Below we lay out the transcript from the live Q&A:

Is it needed to have inhouse data scientists to help deploy AI-powered Automations? 

If that question came to the table a year or two ago, I would say yes, but the beauty of what UiPath is offering is marrying AI and RPA together. 

The same way that RPA has become more off the shelf in the last couple of years, the same has happened with AI. You don't need a data scientist to set up an AI module necessarily or to start looking at AI, the only time that you really need a data scientist, is when you start to get really complicated scenarios, stuff that really needs powerful human cognitive ability.  

If you are using AI to reinforce the power of RPA or for giving the direction of what RPA should and should not be doing, a lot of this stuff is off the shelf. Even a junior developer without leading experience would be able to do that using the different AI Fabric modules that UiPath has. 

And then, the thing is getting a bit of direction from your implementation partner, is it a good idea to use AI for a particular use case? Or can it be tackled with something else? And that's where you should really be bouncing off your ideas in terms of 'should you be using AI in the first place'. 

Have you used RPA to generate all the training data needed for deploying decision making AI models? 

Not all the automation is deployed for decision making models, this requires a little bit of planning, oftentimes we have data, that is where we start.   

But we need to able to turn that data into information that is usable in order to move into knowledge. So in planning for those data decision-making models, we have to start and then push towards knowledge. 

Now lastly, once we have the designers in place that can determine how we would couple Machine Learning and RPA to take action, then we would have intelligent automation or something of the like that would help us, but of course they have to be planned out, and honestly they should be overseen by humans. 

This is that augmentation of the workforce, where we have machines doing the work helping to make the decisions, but at times we would need to ask the human "Are you sure you want to move forward with this?" "Are you giving approval in order to do these things?" and let automation, or machine learning, or additional technologies do their part. 

Does UiPath have preloaded ML models for common automations that require deep learning?  

One of the great tools that is working right now is Document Understanding, and it is a fully-functional machine learning model that as you start to run documents through it, it can off the bat select the data elements out of it and as it learns what those data elements are it's able to apply those to the back-end systems and the more you run the smarter it gets. So there are lots of different use cases, but that is the one that jumps to mind. 

Imagine having to read invoices that are not EDI. You got the invoice in an email attachment, this time was attached in Excel, while this other partner does in a PDF, while other partner does it in the body of the email. As you get Document Understanding up and running, and you train that model, then as it learns the different vendors or customers, or whatever the scenario is, it can readily pick out with great accuracy those data inlets and then apply them to the appropriate systems. 

Should the center of excellence and operating models for AI-driven intelligent automation be managed differently than the traditional RPA CoE 

No, not all, traditional CoE should cover governance support, methodology, operations, model resources, and everything in between. The CoE is really the overarching framework and architecture necessary to scope, design, develop, test, deploy, QA, and measure. So, the CoE should be able to cover advanced technologies, it's just another cog that you put into the machine where the output is efficiency, scalability, or reduction of human error.  

Does RPA integrate easily with any AI technology?  

Easily is a relative term, RPA does integrate with all technologies typically, if there is an API available, if it's web-based or something like that. We can also create custom gateways where needed, but we are seeing less and less of a need for that, as more applications are being built with RPA in mind. 


JOLT is a specialized Robotic Process Automation (RPA) services provider, we can help you and your company navigate the nuances of adapting to a new business reality and setting up an digital-first enterprise.

Contact us to talk to a specialist today about your automation questions and goals, book a free consultation here.

Related content:

Webinar Recordings: