Beyond Big Data: AI/ML Virtual Summit Recap
Vishakh Viswanathan, a Senior Solutions Architect with SDLC Partners, recently presented “Making Bots Smarter: When do you integrate AI into your automation?” for PTC’s Beyond Big Data: AI/ML Virtual Summit. Here we recap the presentation and share food for thought when deciding just how far to take AI in your automation use cases. Or, watch the video on YouTube.
How Smart is Your Bot?
There tends to be two camps – one that wants to use artificial intelligence (AI) to augment human effort and those who want to replace human effort.
Enhancing human capabilities entails mimicking human behavior through technologies like computer vision that completes mundane activities as part of a larger process. It requires uncovering informational patterns so that humans can focus on executive reasoning and decision-making activities.
Replacing human capabilities entails bots that can perform tasks from beginning to end throughout a process. It requires identifying informational patterns that optimize trends so that the bot can make complex decisions, mimicking human cognition.
How is Artificial Intelligence applied to RPA?
AI marries well with Robotic Process Automation because it can complement rules-based automation with cognitive-based technologies like machine learning, natural language processing, and others to create a more hands-off solution that can tackle more complex and dynamic processes and tasks.
AI + RPA Customer Case Study
SDLC Partners created an Intelligent Document Processing (IDP) solution for a national fuels company to streamline their high-volume invoicing process, increase operational efficiency, and reduce mundane tasks to free employees for higher-value initiatives.
When is too much AI not worth the cost and time?
Not every process requires 100% automation. In fact, it might not make financial sense to try to attain 100% automation. In our customer example above, we discovered that their invoicing process included 80-100 types of documents and that 60-70 of those could be templatized, but 30-40 couldn’t and would require an AI solution. Through our analysis, we determined, and achieved, 90% straight-through processing where the last 10% required human involvement. This was the right mix of rules-based and AI-driven automation. The figure below shows the point at which continual AI investments won’t yield enough value and returns for effort start to diminish.
In a future insight, we’ll share with you our approach to evaluating just how much AI maximizes automation returns.
How Should You Maximize the Value of AI in Automation?
Our Intelligent Automation team will help you identify your most valuable use cases on the front-end and guide your automation implementation to achieve the right level of AI value on the back-end.