Generative AI

A Peek at How Automation Anywhere is Implementing Gen AI Internally

  • 20 March 2024
  • 0 replies
  • 58 views
A Peek at How Automation Anywhere is Implementing Gen AI Internally
Userlevel 1
Badge +1

It shouldn’t come as a surprise that, here at Automation Anywhere, we are “customer zero.” If we are going to teach and preach the value of our products, you better believe we’re drinking our own champagne and putting every one of our products to use internally. This is a hugely beneficial setup for our cross-functional teams to provide feedback to the product team on what’s working and what’s not, especially as we develop new roadmap offerings. It’s a wonderful symbiotic relationship that not only helps us bring you the best, tried-and-true tools, but also forms the foundation for us to transform our internal operations.

We formalized our CoE in 2019, and like most organizations getting started with automation, our initial focus was on productivity improvement. This is a great place to start, ripe with opportunities to get your program off the ground and generate positive returns. Fast forward to the present day, we’ve expanded our focus to implement solutions that deliver value in many other ways than just productivity, including cost savings, increased capacity (cost avoidance), and improved customer experience. And like many organizations, we started automating in IT and finance. Now, we have automations spanning from customer service to sales, contract ops, marketing, and more. In all, we were saving 185,000-hours annually through automation. But when a certain hot new technology burst onto the scene, we quickly recognized that we could be generating even more impact…with generative AI.

 

Internal Generative AI Impact

 

Once we had a footing on exactly how we could safely and effectively harness generative AI, we estimated that we would be able to double the impact we had amassed since 2019. That’s double the impact in one-quarter of the time.

Again, we started by implementing generative AI with automations that would amplify productivity of every employee —think meeting transcriptions and a video/article summarizer that any employee in any role could leverage to save time. In parallel, we explored business function-specific applications within more sophisticated processes that bolstered customer service, finance, and even legal efforts.

To understand how great a difference generative AI was making, we conducted an internal study comparing our pre-gen AI use cases and metrics to our present-day gen AI use cases. We found that our “traditional” RPA use cases without generative AI were creating, on average, a 2-3x return on investment. On the other hand, our generative AI use cases were returning 9x (or more) on investment. Triple the return! We documented even more impressive data across business functions, like a 10% CIO budget reduction and a 30% increase in full automated support in areas like IT, customer service, and finance. If we are documenting transformative results like this, there’s no reason you can’t do the same with even the simplest implementations of generative AI.
 

Gen AI Use Cases

 

You know by now that we are not gatekeepers here at Automation Anywhere. We are driven to make our Customers a success, and love to provide resources and share experiences to help you effect valuable change in your own organization. With that in mind, I have two of our generative AI use cases to share with you to hopefully spark inspiration for what’s possible with our products enhanced by generative AI.

Introducing… “Content Assimilator”

I alluded to this use case above, but this was a great launching point for generative AI because it was a fairly simple concept and something that could be made available to impact every single employee. This use case was born from a meeting where, as we were wrapping up, one of my team members highly recommended we read an informative blog article on the topic we had been discussing. We were happy to do so, but when they sent over the link, the article was equivalent to at least ten pages. Frankly, I didn’t have time to read it all in depth. This bred the idea for Content Assimilator.

The process basically takes a link, like a YouTube video or blog article link, passes the content through generative AI, and then sends a summary with key takeaways from the content directly to your inbox. It’s basically your own personal CliffsNotes generator, and we make it available to everyone.


Introducing… “BotScooby”

Every day, our accounts payable team is inundated with inquiries from suppliers. We combined automation with generative AI to streamline an auto-response solution for them and relieve the burden of manually reading and responding to each inquiry, an automation known as “BotScooby.”

Spoiler alert: The most valuable thing that happens here is generative AI classifies and extracts information from semi-structured and unstructured emails. The generative AI classifies inquiries in one of three ways: first, is an auto-response possible? Second, does it require a response at all? Can it just be ignored? This is the case if a supplier simply sends a “thank you” email. We can file those away, but no response is necessary. Third, does this need to be escalated to an accounts payable team member?

This automation is run via remote schedule—every few minutes, it checks for new supplier inquiries. Let’s consider a supplier inquiring about payment status. Here are two scenarios for how this is executed:

  1. Generative AI determines an auto-response is possible with an actual payment status included. It extracts pertinent information, such as the supplier and invoice numbers. Then, the automation queries our payment system via API, which is Workday, retrieves the payment status, and generative AI crafts an automated response with the payment update included. or;
  2. Generative AI determines from the tone of the email that the payment inquiry request needs escalation to an Accounts Payable team member. So, the automation creates a request in Automation Co-Pilot and delivers a notification through Microsoft Teams, one of the applications where the Accounts Payable team typically works. The notification displays a generative AI-generated response for them to edit or approve. They review the request and fire off the response.

With this use case, the Accounts Payable team now only has to spend time on exceptions, and even then, as mentioned in the second instance above, generative AI and Automation Co-Pilot optimize the process to ensure minimal time spent addressing an exception. All the pertinent information is provided, and no one needs to log into separate systems thanks to Automation Co-Pilot.
 

Generative AI + You Next

 

In a recent User Group meeting, we polled our group members on where they were planning to or already implementing generative AI, and over half answered Accounting, with close runners-up in IT and Customer Service. But in all reality, most of our customers fall into the “planning to” phase and have not taken the leap. I get it! There are many considerations with generative AI, and careful planning is required. However, I hope I’ve shed some light on how simple a few generative AI POC use cases can be and how much ROI amplification you can achieve.
 


0 replies

Be the first to reply!

Reply