Community Event

Stellar Keynote Recap: Generative AI Ask Me Anything

Stellar Keynote Recap: Generative AI Ask Me Anything
Userlevel 6
Badge +10

Welcome to our Stellar Keynote Recap Series!


As part of the 1st Annual Pathfinder Community Space Camp & Generative AI Showcase, we’re hosting live sessions with Community MVPs and industry experts to share the latest developments in intelligent automation—especially Generative AI!—and provide you with learnings and resources to drive success at scale. If you don’t have the chance to attend the live session or want to come back to reference some of the critical mission information that was discussed, we’ve captured key intel from each session to share with you!

Day 1 Stellar Keynote covered the spotlight topic of this year’s Space Camp—Generative AI!


We were joined by Peter White, SVP Product, and Pratyush Garikapati, Director, Product Management, to provide valuable knowledge and help answer insightful questions LIVE from our community. This session was truly community-led, with questions directly from our Pathfinder community driving the conversation. Thank you to all our members who contributed such important, thoughtful questions!



We polled our live audience on different aspects of the featured session topic - here is what they had to say!

Poll 1: Have you implemented AI solutions? If yes, what kind of AI solutions have you implemented? If no, what is the top reason you have implemented an AI solution?

  • 60% + of the audience share they have not implemented an AI solution to date.
  • For those that have, the leading solution was Chatbots.
  • For those that have not, the leading reason was Knowledge and Skill Gap followed by Costs.


COMMUNITY TOPIC #1: Implications of Generative AI


💫 Main Intel: Generative AI is the accelerant for intelligent automation. The power of the two things combined—Generative AI & Automation Anywhere—is where the magic happens.

If a process is structured and rule-based, you can use intelligent automation to read inputs and perform a set of actions over your target systems, Salesforce for example. But if a process is not rule-based, one would use a technique like process re-engineering or machine learning to make that business process as rule-based as possible, then apply automation.

What’s changed with Generative AI is the ease with which machines can understand and create unstructured content. GenAI can now extract the content, but generative AI itself cannot access your target system. You still need to take steps to build the entire workflow. That’s where a platform like Automation Anywhere comes in with the tools to update your target systems. And on top of that, the need for governance, auditability, security, compliance, etc. still is entact and may even need to become more stringent.


COMMUNITY TOPIC #2: Automation Anywhere Generative AI Strategy & Roadmap


💫 Main Intel: We want to leverage innovative AI and put that in the hands of our customers to allow them to automate more than what was possible before.

We already have an Open AI package in the Bot Store.

We are one of Google’s early partners for their generative AI solutions, so we will soon be releasing capabilities integrated into Google’s generative AI models. Very soon we will be publishing a Google Vertex Package.

We’re working with Amazon as well.

Our approach with integrations is to be an open system, so you’ll see a lot of options in terms of how you can integrate these models.

Using Generative AI to improve our existing products and capabilities is also in the works to enable our customers to build automations faster and better. More to be announced soon!




Poll 2: What areas of your business could benefit from AI solutions?

56% of our audience selected these top 3 areas:

  1. Customer Service & Support
  2. Finance & Accounting
  3. Supply Chain & Logistics

Other areas that were selected, but not as commonly, were Human Resources & Talent Management, Marketing & Advertising, Sales & Lead Generation, and Manufacturing & Production




Use Case Brief: A contact center with agents is dealing angry customers about flight delays and boarding experience. These complaints flow into a CRM system and in this case we are using Salesforce as the CRM system, but this could be any CRM system. Typically, an agent’s performance is measured on how quickly they are able to turn around a query and produce a response to the customer. There’s also the nature of how well they articulated the response to the customer and whether it actually produced a resolution or more of a reference.

Here we have the agent Zoe interacting with automation through Automation Co-Pilot to address the complaint from the passenger Cameron Steel. We’ve made Automation Co-Pilot available directly in the target application, which in this case is Salesforce.

Solution: Zoe runs a query in Salesforce to determine the relevant data she needs - flight number, departure city, frequent flyer number, etc. - before responding accordingly to the customer. What sits behind this query is an Automation Anywhere automation process. Using the ChatGPT package in the Bot Store, it’s able to extract the relevant information from a blurb of text (in this case frequent flyer number) and classify the query (i.e. flight delay vs lost baggage claim) without writing any code. These 2 capabilities by themselves can create whole slew of use cases.

We have GPT embedded in this demo workflow in a few other places—we also use GPT to respond back to the customer. GPT generates a pre-built response based on the context/nature of Cameron’s complaint, but Zoe can edit the response before she sends it back to the customer.

It’s important to note that the extraction of data from an email or unstructured text has been around for some time with AI models like entity extraction models. The key difference now is you don’t need a data scientist or an ML Ops teams to deploy custom models to do these sorts of tasks. Just using these off-the-shelf models with a prompt, you can instruct it to get data, and that prompt can be as simple or robust as you need and even provide some level of validation or requirement to ensure it extracts the right data, or if there’s an exception, it passes an exception back.


The action shown in the demo is an Open AI Generative AI package in our Bot Store developed by Automation Anywhere. You can download it and use right away.

However, at the moment using these packages requires your own subscription. If you’re using Open AI GPT, you need to have a subscription to Open AI. If you’re using Azure GPT, you need to have an Azure subscription.

We’re exploring alternatives moving forward to make sure the experience of bringing generative AI into your automation workflow is as seamless as possible. More to be announced on that, so stay tuned. But for now you need to bring your own subscription.

COMMUNITY TOPIC #3: Improving Process Discovery with Generative AI


💫 Main Intel: We’re working to accelerate the whole process within our process discovery product to go from observing to automating what you find quicker.

“Chat with your data”—some of the new capabilities in these models around generating code or SQL queries allow you to more quickly visualize and get insights around those data sets you’ve collected.

Then we want to help you turn a discovered process into an automation more automatically so you don’t have to build it from scratch. Stay tuned for more specifics to be announced on what’s coming out and when!


COMMUNITY TOPIC #4: Why Automation Anywhere and Generative AI?


💫 Main Intel: Generative AI + Automation = Business Transformation. It’s not one or the other.

Start by asking—what kind of integration models do you want to expose on top of your existing business or IT landscape?

Technologies like Adept and a few others out there, while promising in nature, are still relatively untested in an enterprise context. What matters at the end of the day, is you still have to understand and incorporate the needs of governance into your existing automations.

Where Automation Anywhere has strength is governance. Time and again we have deployed our product and platform and you’ve seen that scale quite effectively. That’s the assurance an enterprise needs.

It is a co-existence play rather than this-versus-that. You need generative AI to be able to understand what the content is all about and to convert that natural language to systemic instructions, which can then be read through a platform like Automation Anywhere.

If you look at the most powerful generative AI models today, they are not immune to making a lot of errors. You need the workflow and structure that comes from an automation. Then you need to deploy these models strategically at targeted tasks along that workflow where it makes sense.




Poll 3: What is the current, or would be, most important metric for measuring the success of AI solutions for your organization?

  • 70% of our audience answered to Improve Efficiency or Productivity, and Return on Investment


COMMUNITY TOPIC #5: Privacy & Legality Concerns with Generative AI


💫 Main Intel: CHOOSE WISELY and use generative AI models appropriately.

Microsoft Azure’s open AI service, for example, takes none of your data used for model training. It’s controlled within your cloud environment. There is some logging of data that goes into the prompts for abuse monitoring, but even that is an option you can ask to have turned off so there will be no logging of the data.

With Google’s models, which are not broadly released yet, there will be similar assurances around your data—that no data leaves your environment and is not used for retraining general models. You’ll see the same things from Amazon as well.

With the pace and level of investment going into generative AI for businesses, you’ll see a lot of options in the near future with the same level of data, security, and privacy that you would expect from any other enterprise software service.

Re: Healthcare applications

Some of the models already have HIPAA certification. If not, some releasing soon probably will, so there will be options that are compliant for health use cases. Also, we’ve already seen a lot of model variance being built, including smaller models that can run on-prem or in contained environments for those customers who want to do that. We recommend to start building your use cases, not with patient data necessarily, but start building now because you’re going to have some options in terms of technology infrastructure very, very soon.

Also even within health care or other industries that are heavily regulated in nature, there are pockets of use cases that don‘t require you to deal with a lot of data privacy such as revenue cycle management. You can still automate or look at the potential of automating that whole end-to-end business process without actually being forced to divulge privacy-related information.

Peel onion rather than feed the entire vertical!



📌  Don’t look at generative AI as an answer to everything. There is a time and place for it. Understand whether it is the right answer for you and apply it in the right manner. The technology holds a lot of promise and that’s why we’re investing heavily in it, but it does require a fair degree of understanding.

📌  Don’t run blindly with these models. Once you’ve identified a use case and start to roll that out to production, think about how you’re going to monitor that over time. These are statistical models and they have unpredictable behavior in some cases, which is okay and part of what makes them powerful. But you need to monitor them in production. Using your automation workflow to make sure you’re capturing the data and performance around these models is a key part of making them successful.




Announcements are coming in the near future, and as part of that we will make sure our customers have the ability to sign up to expand these conversations further and take a deeper dive around the kind of use cases they see and how we, from an AAI perspective, can really help you build those use cases around generative AI models.

Once we are ready for beta testing, we will invite our customers to sign up for that and they can get early access.

We are looking for feedback on how AAI is making this whole ecosystem available for our customers. At the same time we’d love to hear what kind of use cases or automation scenarios you are looking at in your existing organization.



Our speakers were given an industry and challenged to quickly brainstorm a possible use case to leverage generative AI within that industry. Here’s what they came up with - we only had time for two but look forward to sharing more!

  • Pratyush: Finance - Looking at supply contracts and unearthing terms and conditions in there, then matching those up to the invoices your supplier generates you.
  • Peter: Banking - Generating pitch text for banking and using a combination of capturing information either from the web, specific materials, or reports, then using generative AI to help summarize and extract information from those to construct what would be the outline of your pitch. For example, “Create a summary of this company and financials from their last quarter, extract what their profit was over the last 5 years.”




With the superpower of generative AI on the Automation Success Platform, the universe is yours to explore! What projects will you revolutionize with this innovative technology? Download the OpenAI Generative AI Package today and let your imagination soar!



Looking for additional resources?

Explore our Space Camp Generative AI Special Edition Flight Plan & Tools: 


0 replies

Be the first to reply!