Bob Violino
Contributing writer

Making intelligent automation work at scale

Jul 06, 202312 mins
Artificial IntelligenceBusiness Process ManagementRobotic Process Automation

Enterprise IA pioneers have been blending RPA and AI to great benefits. Here’s how they’ve honed their automation strategies to take IA enterprise-wide.

Female Software Engineer Working in a Modern Monitoring Office with Live Analysis Feed with Charts on a Big Digital Screen. Monitoring Room Big Data Scientists and Managers Sit in Front of Computers.
Credit: Gorodenkoff / Shutterstock

Organizations can reap a range of benefits from deploying automation tools such as robotic process automation (RPA). But adding artificial intelligence (AI) to the mix is where an even bigger payoff can come.

“Organizations have been combining automation and AI technologies for a few years now to improve their business processes,” says Maureen Fleming, program vice president at research firm IDC. “AI tends to broaden the reach and impact of automation, taking on activities that cannot be performed solely through automation.”

Classic examples are the use of AI to capture and convert semi-structured documents such as purchase orders and invoices, Fleming says. “We’re also starting to see NLP [natural language processing] applied to unstructured text, such as categorizing an email or understanding the content of the email,” she says. “Generative AI will significantly and rapidly expand the use of AI to simplify, supplement, and substitute automation.”

Companies that have been using intelligent automation (IA) for a while have learned to leverage this technology at scale, expanding the capabilities to more departments and use cases.

Spreading the news

Telecom provider AT&T began trialing RPA in 2015 to decrease the number of repetitive tasks, such as order entry, for its service delivery group. The group was able to automate one process and then expanded the effort from there, according to Mark Austin, vice president of data science.

Early on in its RPA initiative AT&T decided to combine the technology with data science to create smarter bots that leverage AI capabilities such as optical character recognition (OCR) and NLP. The goal was to create an IA environment to make automation more powerful.

Within a year of the RPA launch, the company had deployed 350 automation bots, and has continued to add more in the years following.

“As part of our intelligent automation program, we now have around 3,000 software bots in production, with about 75 more coming online each month,” Austin says. “What’s more, we’re now reviewing incoming bots to see if we can make them smarter with AI capabilities. We’re finding about 30% of them can be upgraded with AI.”

As demand for RPA spread through the company, AT&T created an automation center of excellence (COE) to accelerate implementation. The COE has made it easier to scale IA by helping develop, deploy, and manage automation efforts throughout the business.

The COE also educates staffers in how to automate various processes. AT&T has trained more than 2,000 RPA developers, who created most of the company’s automation bots. Some lines of businesses have created

their own automation teams, with the COE providing tools and support.

As part of its mission to democratize IA across the company, AT&T is deploying a secure generative AI platform, Austin says. “This is a private instance of GPT-4, which protects our intellectual property from leaking out and gives our employees a secure venue to enhance their productivity,” he says. “We’re equipping this tool with a private ‘knowledge base’ of AT&T-specific data, with chat enabled to get answers directly from these internal AT&T documents and materials.”

GPT-4 (Generative Pre-trained Transformer 4) is a multimodal large language model created by OpenAI, an AI research laboratory.

Since AT&T launched its IA program, “we’ve seen annual benefits of close to $100 million in productivity gains and cost savings,” Austin says. “In a typical year, the return on investment is 10 times.” 

Some of these gains have come from using an AI-powered auto-notification tool that alerts business customers of potential overages based on usage. “This enables our customer care representatives to proactively alert customers before they get hit with higher bills, enabling them to adjust their plans or usage,” Austin says. “We review 21,000 records per minute with this system, which results in happier customers and fewer calls and inquiries to customer care.”

Another AI-powered automation tool AT&T built enables state governments to automatically scan paper documents such as vehicle registration requests, and automate the filing appropriately and store the records in compliance with regulations. Yet another helps large customers move accounts and phone numbers to different parts of their organization seamlessly.

“When the customer calls in, our IVR system accepts the request and triggers a bot that sends a secure web form for the customer to fill out,” Austin says. “Once the form is submitted and the transfer is ready to be made, the bot contacts the customer for final verification.”

Modernizing systems and processes

Investment management provider Capital Group began its automation journey with business process management (BPM), with a goal to digitize manual processes and bring disparate business processes together, says Jim Reis, vice president of technology. This led to a move to RPA for additional automation of highly manual processes within the organization.

“However, after implementing a capability model to assess our current automation investments, we identified gaps in our capabilities that RPA alone didn’t address,” Reis says. “RPA was great for doing unattended work, but there were a number of use cases that required the user to be included for quality assurance purposes. As a result, we quickly moved into more intelligence-driven automation, expanding [our] capabilities” with tools such as intelligent document processing (IDP).

Today, the firm is working with Appian products alongside other technologies to automate end-to-end workflows for its major lines of business, spanning single tasks to complete end-to-end business processes, Reis says.

“For example, IDP uses native AI to quickly and accurately extract data from business documents of all types, for both structured and unstructured data,” Reis says. “This is especially important for us because our work spans many forms of content — from more traditional form-based documents to unstructured email communications.”

Capital Group began investing in IA to add productivity and effectiveness as part of operational scale, Reis says. “By implementing intelligent automation solutions, we’re able to meet client needs at consistent service levels even with fluctuations in work volumes,” he says. “This is something that resonates with most CIOs — understanding how to grow and scale in a healthy way that doesn’t result in increased expenses.”

In addition, the firm deployed IA to drive employee efficiencies. “We recognized inefficiencies in some areas and wanted to utilize our associates’ time and skills for more strategic work, which required freeing them of the tedious, manual tasks that were currently taking up their time,” Reis says.

Another benefit is greater risk management. “Using digitized processes ensures visibility, transparency, and adherence to process, often with service levels and quality assurance steps,” Reis says. “Using automation technologies helps meet client expectations and ensures consistency, while lowering risks that can be attributed to human error.”

Reimaging end-to-end processes

Pharmaceuticals and medical technology provider Johnson & Johnson (J&J) has been using IA for more than three years, with the goal of integrating it within every part of its business. To that end, the company created an enterprise-wide Intelligent Automation Council. Under the guidance of the council J&J is applying IA to support numerous processes.

The company began its automation efforts by using RPA for tasks such as moving documents, completing spreadsheets, and email integrations, and expanded from there into advanced automations. By applying IA to its invoice-to-cash function, J&J increased cash collection, decreased error rates, and cut the number of work hours needed to achieve the same results.

“We continue to make significant progress in operating with a digital-first mindset and reimaging our end-to-end processes with IA,” says Ajay Anand, vice president of strategy and business services for Global Services at J&J.

“We are using insights from our IA maturity assessment efforts to identify large untapped value pools to drive visibility with our executive committee and functional leaders,” Anand says. “In addition, we are also focused on developing a framework for generative AI use case development and prioritization.”

The enterprise IA program is delivering on “experience, effectiveness, and efficiency — giving our employees more time to focus on creative innovations and upskilling,” says Steve Sorensen, vice president of technology services, supply chain, data integration, and reliability engineering at J&J. “It is enabling the reimagining, simplifying, and digitizing processes for employees, patients, healthcare professionals, and other stakeholders, while delivering significant value for the organization.”

For example, the company’s enterprise chatbot JAIDA (J&J Artificial Intelligence Digital Assistant) understands more than 300 intents and is continuously learning through use and user feedback. It frees up contact center employees so they can focus on more complex employee matters, and allows employees to focus on more meaningful work, Sorensen says.

The company is also using digital twins in manufacturing to unlock new capabilities for product innovation and consumer engagement. “Digital twins platforms create digital replicas in virtual reality that mimic the physical supply chain, which can then be modified into different scenarios to optimize product flow, maximize efficiencies, and minimize costs,” Sorensen says.

IA at scale — tips for success

Experts weighed in with tips on how to successfully use IA at scale.

Know your needs and capabilities. It’s a good idea to first know the organization’s needs and capabilities before making the necessary investments in tools, Capital Group’s Reis says. “Every business challenge is different, so having a formal assessment in place is critical for understanding what kind of automation technology is needed or if you can leverage an existing capability to solve the issue,” he says.

This is also a critical step for ensuring buy-in from key stakeholders and executives and the teams that will actually be using the technology, Reis says.

Roadmap for benefits, not tech. IDC’s Fleming advises creating a roadmap aimed at achieving demonstrable benefits using the appropriate automation and AI tools for each use case, rather than focusing on technologies, Fleming says. “Scaling automation shifts from tactical one-off efforts to a strategy to improve one or more business processes,” she says.

This typically involves using discovery tools to explicitly understand where improvements across a business process need to be made to fix the inefficiencies, Fleming says. “The statistics in discovery create a scope of the problem and how each issue can be solved, whether by business redefining their process or by applying technology,” she says.

Secure sponsorship. Be sure to establish senior-level enterprise sponsorship for IA, J&J’s Anand says. “It’s important to enroll the C-suite and executive committee for support and sponsorship upfront,” he says. “By ensuring that they understand the 3E [experience, effectiveness, and efficiency] value-creation potential, the digital transformation will have champions at the highest levels.”

Pilot to accelerate results. Another good practice is to test and learn from solutions early and often. “Using proof of concepts or pilots allows you to get real-world, business-tested results — and quickly,” Reis says. “Even after assessing your needs and finding a solution that fits those needs, it might not end up being the right solution in practice. This is especially important with intelligent automation technologies as they vary widely from vendor to vendor.”

Continuously survey the IA landscape. Stay current on the latest technologies because the industry is moving quickly, Reis says. “One of the ways we do this at Capital Group is by doing an assessment of the industry, which we call a landscape survey,” he says. “We look at the industry every 18 months and dive into who the market players are, and we do this in conjunction with other research firms as well.”

Consider a CoE. It might make sense to start small with an IA project and scale up by involving the entire organization, AT&T’s Austin says. “A center of excellence that’s centrally funded can help people who aren’t data scientists get up to speed,” he says. “Of our 3,000-plus bots, 92% of them are built in the business units, not the Chief Data Office.”

Track, measure, and reuse. It’s also a good practice to deploy a platform to track how the bots are functioning and their uptime, Austin says. “You also want to make bots as reusable as possible, so others can plug them in to their operations,” he says. “In addition, automate the approval process to remove bottlenecks, and have senior executives such as the chief financial officer or chief technology officer evangelize for these tools.”

Automate AI enablement. AT&T’s Austin also advises putting automation to work within your IA strategy. Here, generative AI may be key. “Develop an automated capability to encourage and recommend AI-enablement for your bots,” Austin says. “Generative AI is proving to be a massive benefit and way to increase the value of our bots.”