What's the next big thing in enterprise automation? According to the tech giants, it's agents, powered by generative AI.
Although there is no universally accepted definition of an agent, the term has recently been used to describe AI-powered generative tools that can perform complex tasks through human-like interaction between software and web platforms. I am.
For example, an agent can create an itinerary by entering customer information on an airline or hotel chain's website. Alternatively, agents can automatically compare prices between apps and order the cheapest ride to your location.
Vendors sense an opportunity. ChatGPT maker OpenAI is reportedly deep into developing an AI agent system. And Google demoed a number of agent-like products at its annual Cloud Next conference in early April.
Analysts at Boston Consulting Group recently cited experts who predict autonomous agents will become mainstream within three to five years, saying, “Enterprises can start preparing for mass adoption of autonomous agents today.'' “There is a need,” the report says.
old-school automation
So where does that leave RPA?
Robotic process automation (RPA) became popular more than a decade ago when companies began turning to the technology to enhance their digital transformation efforts while reducing costs. Like agents, RPA drives workflow automation. However, this is a much more rigorous form, based on “if-then” preset rules for processes that can be broken down into tightly defined discretized steps.
“RPA can mimic human actions such as clicking, typing, and copying and pasting to perform tasks faster and more accurately than humans,” Saikat Ray, VP analyst at Gartner, said in an interview with TechCrunch. explained. “However, RPA bots have limitations when handling complex, creative or dynamic tasks that require natural language processing and reasoning skills.”
This flexibility makes RPA expensive to build and severely limits its applicability.
According to a 2022 study from RPA vendor Robocorp, 69% of organizations with RPA experience an automation workflow failure at least once a week, and many fail to fix it. It's been several hours. The entire business consists of helping companies manage their RPA installations and prevent corruption.
RPA vendors are not naive. They are well aware of the challenges and believe that generative AI can solve many of them without hastening the demise of the platform. In the minds of RPA vendors, RPA and generative AI-powered agents can coexist peacefully, perhaps even growing to complement each other one day.
Generative AI Automation
UiPath, one of the leading companies in the RPA market with an estimated 10,000+ customers including Uber, Xerox, and CrowdStrike, recently announced new generative AI capabilities focused on document and message processing and UiPath CEO Bob announced the execution of automatic actions provided by Mr. Enslin calls it “one-click digital transformation.”
“These capabilities provide customers with generative AI models trained for specific tasks,” Enslin told TechCrunch. “Our generative AI powers workloads such as email text completion, classification, image detection, language translation, and the ability to filter personally identifiable information. [and] Instantly answer questions related to people topics based on knowledge from internal data. ”
One of UiPath's recent explorations in the generative AI domain is Clipboard AI. It combines UiPath's platform with third-party models from his OpenAI, Google, and others to, in Enslin's words, “bring the power of automation to anyone who needs to copy/paste.” ” Clipboard AI allows users to highlight data in a form, leverage generated AI to determine the appropriate destination for copied data, and point to another form, app, spreadsheet, or database. can do.
Image credit: UiPath
“UiPath recognizes the need to marry action with AI. This is where the value comes from,” Enslin said. “We believe the best performance comes from a combination of generative AI and human judgment, so-called human participation, throughout the end-to-end process.”
UiPath's main competitor, Automation Anywhere, is also looking to incorporate generative AI into its RPA technology.
Last year, Automation Anywhere created workflows from natural language, summarized content, extracted data from documents, and perhaps most importantly, adapted to app changes that would normally cause RPA automation to fail. We have released a generation tool that utilizes AI.
“[Our generative AI models are] developed on [open] Peter White, senior vice president of enterprise AI and automation at Automation Anywhere, told TechCrunch. “We continue to build custom machine learning models for specific tasks within our platform, and are currently using automated datasets to build customized models on top of our foundational generative AI models. Masu.”
Next generation RPA
Ray says that as generative AI contributes to increasing RPA capabilities, it is important to recognize its limitations – biases and illusions. But risks aside, he believes generative AI will transform the way these platforms operate, adding value to RPA by “creating new possibilities for automation.”
“Generative AI is a powerful technology that can enhance the capabilities of RPA platforms to understand and generate natural language, automate content creation, improve decision-making, and even generate code,” said Ray. . “By integrating generative AI models, RPA platforms can deliver more value to customers, improve productivity and efficiency, and expand use cases and applications.”
Craig Le Clair, principal analyst at Forrester, sees RPA platforms as ripe for expansion to support autonomous agents and generative AI as use cases expand. In fact, he expects RPA platforms to transform into an all-purpose toolset for automation, a set of tools that, in addition to related generative AI technologies, will support RPA implementation.
“RPA platforms have architectures to manage the automation of thousands of tasks, which bodes well for centralized management of AI agents,” he said. “Thousands of companies have well-established RPA platforms and will be willing to use them for AI-infused generative agents. RPA easily integrates with existing work patterns through UI integration. It has grown because of the capabilities it can provide, and this will continue to be valuable for more intelligent agents.”
UiPath has already started working in this direction with a new feature, Context Grounding, which began previewing earlier this month. As Enslin explained, Context Grounding transforms business data consumed by generative AI models (both first-party and third-party) into an “optimized” format that is easier to index and search. It is designed to improve the accuracy of the model.
“Context grounding extracts information from company-specific datasets, such as knowledge bases and internal policies and procedures, to create more accurate and insightful responses,” said Ensslin.
If there's one thing holding back RPA vendors, Le Clair says, it's the ever-present temptation to lock in customers. He emphasized the need for platforms to “remain agnostic” and provide tools that can be configured to work with a variety of current and future enterprise systems and workflows.
In response, Enslin promised that UiPath will remain “open, flexible and responsible.”
“The future of AI will require a combination of specialized and generative AI,” he continued. “We want our customers to be able to use all types of AI with confidence.”
White was not strictly committed to neutrality. But he emphasized that Automation Anywhere's roadmap is largely shaped by customer feedback.
“What we're hearing from customers across all industries is that generative AI has dramatically increased our ability to incorporate automation into more use cases,” he said. “We believe that by incorporating generative AI with intelligent automation technologies like RPA, organizations have the potential to reduce operational costs and increase productivity. “We will have a hard time competing with other companies that are adopting automation.”