Artificial intelligence is coming to enterprise software near you, and it will push both IT pros and end users out of their comfort zones.
As machine learning, bots and analytics capabilities mature, more software providers are ramping up the use of artificial intelligence (AI) in their products – including those for IT departments seeking better ways to handle large volumes of data, automate processes and improve end-user productivity. The technology is in its early days, but some AI tools are lending themselves to the citizen technologist by emphasizing simplistic management rather than the complexity often present in emerging technology.
"I see [machine learning] working best to present suggestions and outcomes to a user, who can then drill down into information and approve or change those suggestions," said Adam Fowler, IT operations manager at a law firm in Australia. "There are some workflows that won't need this of course, but I believe the combination of AI and user gives the best outcome."
Automation for the people
AI and machine learning can provide a boost to many types of software. Organizations that rely on common, routine workflows can benefit from automation tools that use AI to learn workflow patterns and alert users to tasks to improve their efficiency, for instance.
AI tools are useful in combination with automation workflows to help users make decisions about what action to take next, what information to present, or what content -- such as an email -- to generate from that information, Fowler said.
"I see AI, just like automation, taking away the repetitive tasks that people do today," he said. "If you have to analyze data and make a choice, AI can create huge efficiencies in those processes."
Fowler's firm isn't using much AI yet, but it can be very useful in the legal industry, which requires users to weed through hundreds of thousands of documents to determine what information is relevant to a case, he said. In that scenario, AI in automation tools can help locate what's relevant using keywords that a user enters, and spit back analysis of the data it finds.
"You can then start with the most relevant documents and potentially save a huge amount of time for yourselves and your clients," Fowler said.
One AI-based offering comes from Nintex, a global process automation software provider based in Bellevue, Wash. Typical use cases for its automation tools include sales departments and call centers, which require collaboration among team members or departments, and use standard applications such as mobile forms, chief marketing officer Matt Fleckenstein said.
Amazon, Microsoft, Google and IBM are the major players in creating machine learning capabilities for automation software. Then, smaller providers such as Nintex, for example, can take advantage of these capabilities. The company's Hawkeye product looks at the processes users perform in their applications and automatically flags tasks that need their attention, using machine learning technology to understand what specific responsibilities a user has.
The language of analytics
Another area where AI is ripe for growth is analytics, which aims to gather, evaluate and provide feedback on data. Machine learning can help take analytics one step further by identifying patterns of data and understanding over time a users' custom preferences for how that data is presented to them, to offer the user more targeted results and automatically take actions for them.
Apogee Legal, a legal analytics firm in Charlotte, N.C., helps large corporations implement AI tools that analyze their legal contracts to address regulatory compliance issues, gather information related to vendor and customer obligations and more. Sometimes, the information that a client needs to locate across its contracts is straightforward -- what type of contract it is, what dates are on it, when it expires -- but often it's more complex.
A common use for AI-based analytics among Apogee Legal's clients is to assess whether a contract complies with the upcoming General Data Protection Regulation (GDPR). AI tools can determine if contracts address the concept of the duty to notify authorities in case of a data breach, which is a GDPR requirement, said Jim Wagner, managing director at Apogee Legal.
This task is especially difficult for contracts that were written many years ago, and across contracts that use varying language, he said. Natural language processing (NLP) in AI tools, however, can overcome the challenges that people cause by using different terms and types of writing and language.
This machine learning process uses NLP to find the topics a user needs to review within the contract; for instance, something might be called '"governing law" and later in a contract be called "applicable law," but the AI can understand that they refer to the same topic and find both when needed.
Test your knowledge of AI and machine learning features
Machine learning can help businesses gain powerful analytics value from their data -- but only if it's done right. How much do you know about machine learning and related forms of AI?
"When you think about analyzing tens of thousands of contracts … to do that through a manual process is an overwhelming effort," Wagner said. "The AI tools allow us to analyze a wide variety of contracts with a wide variety of language used … and give clients visibility into what would otherwise be the dark matter of contracts."
Apogee Legal's client CyrusOne, a global data center provider based in Dallas, uses a tool called Seal Software to extract terms from contracts and load them into a database to analyze.
"This allows our legal department to quickly analyze and evaluate large number of contracts in different scenarios," CyrusOne CIO Blake Hankins said. "AI helps us save time and money by allowing us to employ AI resources instead of employee or contractor resources. This … significantly cuts the time needed to get the work accomplished."
AI for IT administration
AI is also coming to the IT management and security products that administrators use to manage end users' resources. Security tools, for instance, might use machine learning to learn over time what risks are present for certain networks, devices or applications, and identify them before they can cause a breach. Application performance management is a big use case too, as these monitoring tools glean more data than ever about the apps they're tracking, said Michael Azoff, a principal analyst at Ovum, a technology research firm in the U.K.
AI helps IT handle the massive amounts of data that these tools create, he said. In particular, AI-based automation can help organizations manage that big data from its existing software systems at a more granular level, he added.
"IT departments that want to use AI need to have a regime for data management -- extraction, transformation loading, how they manage their big data," Azoff said. "Do they need data lakes? That's really critical for feeding into the AI systems."
Getting users in the AI mix
One commonality among emerging AI tools is an effort to make them easy enough for non-tech experts, or citizen technologists, to take advantage of. It doesn't mean those users are creating the machine learning algorithms, but they're able to pick and choose which AI capabilities they want to use in their workflows and customize them.
With Nintex, for example, an employee can create an AI-powered workflow that automatically sends a contract to the appropriate legal department employee for approval, based on what's in the document, the time of day and other factors. Nintex also includes management and analytics features that give IT visibility into how people are using these AI capabilities.
Because machine learning is essentially making decisions for users, IT departments still need to ensure that they can maintain oversight over what the AI is doing.
"That's a big issue, which is why I believe AI and users need to work together [with] checks and reviews to make sure AI is doing exactly what you want," Fowler said. "I can foresee people expecting exact results, like a calculator; however, the data being looked at is contextual. We generally see issues when too much AI and automation is implemented."
The IT department at CyrusOne made sure to test its tools' AI capabilities to analyze the results it returned before letting the software execute any tasks automatically, Hankins said.
"AI is helping us in analyzing data to assist in making decisions," he said. "AI can work well in most situations, but testing and ensuring success before turning automated AI decisioning on is key."
Michael Azoffprincipal analyst, Ovum
Seal Software comes with a pre-built set of AI capabilities that help users find basic information in contracts, such as job titles and common laws. It also lets users train the software to home in on information that is most relevant to their needs.
For instance, a unique use case might be a user who wants to find information about requirements for background checks in contracts, Wagner said. Seal allows the user to teach the software how to find that information so it can do so going forward.
"People who are not technologists can execute the machine learning training," Wagner said. "You have to have a comfort zone with dealing with these tools in order to train them. You'll see a mix of people with technology background and with exposure to the subject matter who are doing the training."
This type of capability may be especially useful given that IT expertise in AI is somewhat green. In the U.K., for instance, there were 2.3 job openings for every qualified candidate in the area of AI and machine learning software developers as of October 2017, according to Indeed.com data in a ComputerWeekly.com article.
"People are still learning how to use this technology," Azoff said. "There's a huge skills gap. From what I see, companies are struggling to find the right people to be able to exploit this knowledge."
Organizations should expect to see growth in the IT job market around AI development and management skills as the technology matures, however, he said.
The share of Indeed.com job listings requiring AI skills has increased by 4.5 times since 2013, according to Forbes analysis of the site's data. And machine learning trainer/scientist and AI developer are two IT job roles to watch in 2018, according to CompTIA.
For now, the key to adopting AI tools in the enterprise is ensuring that these machine learning capabilities meet the goals of the organization and line up with the challenges it's facing, said Miriam Deasy, senior analyst at Ovum.
"AI techniques will become so common as time goes forward," she said. "The real challenge for IT pros is to figure out how they can add value with these recent developments. How can this technology be woven together? There's still a lot of educating to be done."
- Resolving key integration challenges for financial applications –SearchFinancialApplications
- The Right Strategy for CFOs to Navigate Financial Application Purchases –SearchFinancialApplications
- Sound application development rests with secure APIs in the cloud –SearchCloudApplications
- Financial Reporting and Analysis: Derive Meaning from Financial Data to Improve... –SearchFinancialApplications