Artificial Intelligence
Oct 09, 2017

AI, automation meet reality - in 5 steps

Four out of five companies say that they should get more from digital. The cause? A confusing, over-hyped, heterogeneous landscape of point solutions. The analysis of hundreds of organizations and thousands of operations running new digital solutions reveals that five steps can help businesses harness digital effectively.

Give shape to data...

AI enables organizations to manage their enormous volumes of information and unstructured data, making it easy for other programs and applications to use.

Imagine 10 different invoices being fed into an ERP from 10 different channels, and all of them being handled as though they’d been read by human eyes. Human eyes that don’t make human errors. Or imagine applications at an insurance company that process vehicle damage claims and images in bulk. Two real life examples:

  • An industrial manufacturer uses natural language processing and advanced analytics to audit its logistics contracts for the sake of efficiency. It goes on to cut $40 million from a total logistics spend of $600 million.
  • A hospitality company uses natural language processing and smart optical character recognition to speed up account reconciliation, with a 43% improvement as a result.

... to make robots work harder...

With standardized data, robotic process automation (RPA) and other automation tools can significantly enhance, and often exceed, human performance across more processes and areas of the business.

Traditionally, robots can't handle change. Even the slightest discrepancy in inputs sends them spinning. Which means that people have needed to intervene, arduously sorting and preparing data. Resulting in bottlenecks, errors, and poorly performing automation.

But with data conveniently tidied up and formatted by AI, the bots can quickly and confidently get work done and take more on. And it’s not just robotics that sees the benefits of greater speed and accuracy. Systems of engagement – thin layers of cloud-based software that sit on top of older systems of record or ERPs – are another place we see AI-enabled automation making an impact. For instance:

  • A Fortune 20 insurer has used RPA to reduce manual work by 48% and costs by 40% over a five-year period. All while achieving 100% accuracy. 
  • A global aerospace giant implemented a center of excellence to drive RPA, and expects savings of $490 million over the next five years.
  • A multinational food and beverage conglomerate has seen $53 million in cost savings over the past five years from touchless procure-to-pay processing. Robots now process 78% of invoices and 65% of helpdesk queries.

... to create sharper performance insights...

Instead of wasting precious time struggling with too much data and not enough useful information, business decision-makers can leverage prescriptive analytics to model scenarios faster and more precisely. They can then analyze results in real-time to drive better business outcomes. AI and advanced machine learning algorithms help by sifting through data to find inferences or patterns – factors that could affect processes or business performance. For example:

  • A healthcare company has eliminated duplicate invoices, and has improved visibility and control of its operations and supplier relations by using predictive analytics. It has also seen run-rate savings of $30 million.
  • A pharma manufacturer has reduced costs by 20% by using predictive analytics to improve its e-invoicing capabilities.
  • A food and beverage company now enjoys a uniform, real-time view of its data across all regions and functions. Thanks to analytics and advanced visualization it has also improved the end-client experience.
  • A global consumer packaged goods company wasted weeks trying to interpret both structured and unstructured data from internal and external systems for financial reporting. Using an AI-based reporting tool, the business now generates reports in just a few days, automating 70% of data collection.

...then channel remaining work between people and machines...

While AI and robotics may be doing much of the hard work, this doesn't mean the people in the organization can put their feet up - machines will not be able to "finish the job" for some time. Be it exceptions or intractable cases requiring human expertise or intuition, people will still be needed. Dynamic workflow tools allocate work to the right human or robot, at the right time.

Tasks that require nuance, empathy, subtlety, and the ability to infer complex connections still need the human touch. Prospect inquiry handling is a great example. Hot leads need to be handled as a priority, but more complex cases are best left in expert hands.

Sophisticated, dynamic workflows can now make intelligent decisions about work allocation. They can escalate the right tasks, to the right people, in the right part of the world, at the right time. Three cases where this was the path

  • A French pharmaceutical company orchestrated its SAP systems with dynamic workflow. Now it can respond 40% faster to more than 800 customer calls a week.
  • The largest automotive retailer in the United States transformed business processes, and saved more than $500,000 a year in labor costs.
  • A hospitality company uses dynamic workflow and robotic process automation to increase cash transaction speed by 36%, with 100% accuracy.

...finally, keep improving processes with machine learning.

Machine learning engines can learn from and automatically improve the results of these cycles - where input, output and "how things were done in between" are now explicit and can allow machines to learn.

And based on this information, it can make proactive, educated decisions that improve operations. Like automating relatively simple process orders or moving a low-impact task to a more junior employee.

This approach ensures your most valuable (and expensive) employees only work on the most complex, high-priority cases. And that you deliver faster turnaround, better customer experiences, and more cost-effective operations. Especially when it’s crunch time. For instance:

  • A top pharmaceutical company is using AI, analytics, and predictive modeling with other technologies to transform how the company tracks, predicts, and solves drug-safety issues–resulting in saved lives, 100% compliance, and cost reductions.
  • A global consumer products manufacturer applies design thinking, conversational AI, and analytics to reimagine order management. The result: cut stock-outs by half to improve the consumer's experience.

Five steps powered by machines but designed for humans.

While organizations and processes differ widely, these five steps are very commonly appropriate for a large variety of cases. They can offer a simple narrative to engage management and employees alike in creating a shared vision and identify a set of potential priority initiatives.

What's more, they can prevent software vendors and service providers to hijack the conversation towards what they have - be it machines or people - instead of solving the real equation of our time: how to design machines and processes to give people 21st century jobs.

About the author

Gianni Giacomelli

Gianni Giacomelli

Chief Innovation Leader

Gianni serves as Chief Innovation Leader where he drives and sponsors Genpact’s strategic initiatives aimed at sustaining clients’ transformation into digitally-enabled companies. He also co-leads the Massachusetts Institute of Technology (MIT) efforts to set up a Collective Intelligence Design Lab.