Chief Content Officer,
Techstrong Group

Automation Anywhere this week extended its robotic process automation (RPA) platform for processing documents to include a set of generative artificial intelligence (AI) agents designed to learn and adapt as processes evolve.

Launched at its Imagine 2024 conference, the AI + Automation Enterprise System is based on a second-generation GenAI Process Model that has been trained using metadata from more than 300 million process automations running on the Automation Anywhere platform.

Capabilities added to that core Automator AI lifecycle management platform include an Enhance Autopilot tool for automating workflows across cross-functionals, and a Generative Recorder tool to detect changes to user interfaces and, if needed, automatically remediate issues.

At the same time, the Document Automation service the company provides now also leverages generative AI to process in real-time any document type, including unstructured documents, with 90% accuracy.

A low-code AI Agent Studio tool has also been added that leverages foundational generative AI models from Amazon Web Services (AWS), Google Cloud and Microsoft Azure OpenAI Service, including now the Azure AI Document Intelligence service, to create agents capable of automating a specific task in a way that doesn’t require any data science expertise.

Automation Anywhere is also previewing Automation Co-Pilot, a generative AI tool that invokes the Amazon Q service to enable organizations to automate workflows in addition to making available a set of pre-trained Service Operations Solution Accelerator agents for automating common tasks.

Organizations will also be able to customize AI agents using a retrieval-augmented generation (RAG) service that Automation Anywhere has made available or, starting this fall, the AI modes made available on the Amazon service.

Additionally, Automation Anywhere will also provide a set of embedded prompt testing, monitoring, compliance, auditing, security and governance capabilities to ensure outputs are relevant before any agent is deployed. In the near future, Automation Anywhere is also committing to providing tools to mask sensitive data.

These agents differ from existing RPA bots in that they are capable of learning versus repeatedly executing the exact same rote task, says Claudia Michon, senior vice president for product and solutions marketing for Automation Anywhere. “RPA is more of a rules-based approach,” she notes.

The goal is not to replace RPA bots, but rather provide organizations with a single platform through which they can mix and match both as needed, adds Michon.

It’s not clear how the relationship between RPA and generative AI will evolve, but it’s becoming increasingly clear that the level of skill required to build and deploy an AI agent is rapidly declining. Many Digital CxOs will soon be able to use AI to automate a wider range of tasks without necessarily always having to tap into the expertise of data scientists that are hard to find and retain.

The one thing that is certain is the pace at which processes can be automated is about to significantly accelerate. In fact, rather than always having to advocate for change, the next major challenge Digital CxOs might face in determining how best to orchestrate a wave of digital business transformation enabled by generative AI tools that are readily available to anyone in the organization.