The market for robotic process automation (RPA) has grown rapidly over the last decade with adoption often driven by business users excited by the opportunity to achieve cost savings and process improvement without the usual headaches of big technology projects. Despite this popularity with business users, many IT professionals have derided RPA as being no different from macros or merely a Band-Aid that would have a short lifespan in the enterprise.

But RPA can be more. As the category matures, it will prove to be both a gateway for developing practical applications of AI in the enterprise and a tool for digital transformation initiatives. And we saw that yesterday and today at leading RPA vendor UIPath’s annual conference in Las Vegas: UIPath is making big strides in delivering on these two goals, moving well beyond RPA’s early bot-centric model for automating tasks and pointing toward a future where RPA is a robust platform for building applications that automate complete business processes.

The company is adding a new framework called AI Fabric, and it is also making it substantially easier to bring machine learning into automations.

It’s worth stepping back for a moment to assess where this has all come from and why Gartner reports that RPA is “…the fastest-growing segment of the global enterprise software market.”

The nomenclature of “robotic process automation” is less than a decade old and refers to the core capability of the software: to emulate the mouse clicks and keystrokes that a human would do, thus automatically retrieving or entering information into computer systems.

This capability can have a profound impact on an organization’s costs. Computerization of business has often introduced new manual tasks that requires very little human judgement but a high volume of clicks and keystrokes, and RPA can provide substantial savings for organizations by replacing the human in these tasks.

But as companies began implementing this technology to eliminate rote and repetitive work, RPA’S limitations became evident: First, the tasks RPA automates sit within larger business processes where work requires coordination between multiple people or functions. Second, decisions are often required in the midst of these processes. For example, in processing an invoice from a vendor, the finance department may need someone to approve the invoice before scheduling payment. RPA might automate entering the information into a system, eliminating that rote work. But it hasn’t been able to route and manage the approval request process.

With its announcements this week, UIPath has now reinvented RPA, announcing a vision to provide organizations with a complete application platform. In this new way of thinking about RPA, organizations will build automation applications that manage the interaction of people and bots in processes. In addition, the introduction of AI Fabric makes it substantially easier for organizations to use machine learning to make decisions and judgements and incorporate these into automations.

In an example presented at the conference, a financial institution had developed a machine learning algorithm for credit review and approval. Using UIPath’s AI Fabric, the company was able to easily add this automated decision into a loan application review process.

Transforming RPA into a “platform” will substantially increase the places where enterprises can use automation technology. Virtually everything a company does can be transformed by having tools that manage work and eliminate activities that can be automated. While many of the traditional enterprise software vendors will add automation capabilities to their applications, enterprises will still be saddled with the challenge of having multiple disparate systems and functions that need to be connected. An automation application platform will give them a way of managing these connections.

Imagine the typical “where is my stuff” customer service inquiry reinvented using this new automation application framework. Integration between call center software, CRM systems, and the system a company uses for tracking shipments could allow for an automated message to be played right when a customer calls in, reporting the status of a shipment and why it is late before the customer ever reaches an agent. And if the customer still wants to speak with an agent, a machine learning algorithm might advise the agent on the right offer to make to the customer based on history, size, and frequency of purchases, so the customer remains loyal and satisfied.

Over the next decade, every company will rethink all work — what gets done, by whom, and when to automate or apply AI. UIPath’s new capabilities are exciting because they’re making brand new options possible.

[VentureBeat regularly publishes guest posts from experts who can provide unique and useful perspectives to our readers on news, trends, emerging technologies, and other areas of interest related to tech innovation.]

Ted Shelton is founder and CEO of Robodomo and has been working in the field of automation and AI for almost a decade.