|
by Andy Lientz on May 17, 2018 Not long ago, enterprise IT departments would purchase a new application, deploy it to thousands of employees and hope for the best that they would use it. If the app didn’t align precisely with workflows, or if valuable features were hard to find, there was a good chance usage would be low. We’ve come a long way since then. Interfaces for business software have greatly improved and some applications allow a degree of customization to match the task at hand. But there’s a lot more that can be done to ensure enterprises get maximum value and productivity from their enterprise applications. Machine learning will make a lot of that possible.  In the cloud, it’s relatively easy for application providers to collect data about how software is being used. Algorithms can then be applied to uncover patterns and deliver them to users with improved recommendations and interfaces. It’s important when designing these systems to adhere to strict data privacy and compliance policies, which prevent us from using personal or company specific data. However, there are still valuable patterns to uncover in how software is configured and used. Here are four ways we can apply machine learning to make business software smarter, improve engagement, and recover valuable hours in the day. 1. Data ConsistencyFor a team, ensuring data is collected in a consistent manner is vital for businesses seeking to leverage that data to improve operations, but enforcing consistency across a large organization is a challenge. There is a lot of time spent “cleaning” data manually to make information roll up or to report out on what is working and what’s not working. For example, take a simple form that requires a task, a date, and a status to be filled in when a task is completed. If the date is left blank, the data from that form is misleading or not useful. If we start applying machine learning to the data entry patterns, we can have the system generate an alert when there’s inconsistent data, such as something marked “done” but with no end date or something marked “everything is good in status” when it’s overdue. We can then flag it to the employee, who gives a thumbs up or thumbs down that trains our machine learning algorithm to help the organization be more consistent with data. 2. EngagementA lot of business software includes features that are highly useful but go unused because the employee simply doesn’t know they’re there or that they may be useful. Through machine learning, we can observe patterns of behavior of how tasks are completed, and recommend that pattern when other users start to perform the task in the future. For instance, if someone is copying and pasting a number of rows into Smartsheet it may be they don’t know about the forms feature, and are manually entering in responses from email. By letting the user knows the feature exists, we can expose a feature in Smartsheet and make the user more productive. 3. Natural Language InterfacesSometimes, speaking or typing commands using natural language is a far quicker way to get work done, and advances in machine learning have made this technology practical for business use. At Smartsheet, for example, we’ve started incorporating Converse.AI Chatflow technology, which applies natural language technology to automate workflows and data exchanges. We recently released Smartsheet for Workplace by Facebook, which uses a natural language interface in Workplace Chat to set up the integration and enable users to receive notifications, update requests, and approval requests through chat.  In the future we see this type of bot interaction being extended to other chat platforms and speech interfaces. Natural language has tremendous potential to make software more intuitive. 4. Advanced Process AutomationAdvances in this technology also has the potential to automate even more complex tasks than we do already, even to the point of automating entire workflows. For example, if an employee in marketing is launching a new campaign, an application could build a workflow and start populating fields with the people and organizations the marketer typically contacts for that type of program. In this way, machine learning would enable us to take a process and map it to the software to perform a workflow — without the employee taking any action at all. Transforming the Future of ITSome of these changes may seem minor, but amplified across a large organization they add up to many hours saved — hours that can be spent on work that is far more valuable to the business.
Historically, customizing software in this way has been a slow process that tied up the IT department, and it couldn’t be justified unless a process was highly repeatable across many workers. Now we can deliver these improvements at the level of the individual, with no overhead for the customer at the organizational level. Ultimately, machine learning has the potential to reclaim hours a day for our users. Source: Smartsheet Blog Comments are closed.
|
AuthorWriters and Bloggers from Smartsheet. Archives
July 2018
Categories
All
|
RSS Feed