As new business processes arise in the face of shifting pandemic realities, Business Technology (BT) professionals have to find new ways to adapt and automate across all levels. To help with these automations, BT professionals are increasingly embracing the AI-assisted possibilities of machine learning (ML), in which computing technology is utilized to extract insights and solve complex business problems with minimal human involvement. 

According to FlatWorld Solutions, the benefits of ML are multiple, from “enhancing business scalability and improving business operations” to predicting customer behaviors and beyond. 

To help explain the various ways that ML can be utilized to assist with internal and external communication as well as improving customer experience, Kumad Kokal of Stitch Fix and Pranav Shahi of Atlassian laid out a number of use cases within their own companies at Biz Systems Magic 2021.

The Stitch Fix Story: ML As Self-Service Support

As the Senior Director of Enterprise Applications and Insight at Stitch Fix, Kumad Kokal found that the company’s IT support staff had become overwhelmed with too many tickets and a workforce that was spread across a number of different localities and time zones. To help mitigate this, Kokal looked to ML-assisted tools to transform the ticketing process into one where more problems could be easily solved through self-service.

Kokal laid out five areas of focus when approaching this problem: building an AI/ML-based task resolution tool; fast-tracking its implementation to relieve COVID-related stresses; situating this tool within employees’ natural workspace—in this case, Slack; making sure that the self-service application is available globally at all times; and ensuring that the tool would integrate easily with other apps including messaging, SAML service, ticketing tools, and more. 

The result was Astra, an AI Chatbot that helps guide employees through IT-related issues. Some features of Astra include feedback buttons to ensure the ML process can be further honed and perfected, seamless integration with other apps, and even the ability to be proactive and notify employees about things like upcoming contract and password expirations. 

Astra has helped resolve 2,000 tickets, according to Kokal, with an average of 3.5 tickets per day, taking a tremendous workload off of support staff and allowing them to focus on more heavy-lift assistance. With the expansive possibilities of adaptive ML, Kokal hopes that Astra can expand to cover API integrations and automated user provisioning, as well as provide finance and payroll support.

The Atlassian Story: ML As Tool for Advanced Analytic Processes

While Kokal and Stitch Fix utilized ML as a way to optimize internal communications and support, the technology can also be used to assist with complex analytic processes such as topic modelling and sentiment analysis, as well as processing and archiving documents. 

At Atlassian, Pranav Shahi considered how ML could be used for such processes. Before expanding on the company’s various use cases, Shahi first explained the three forms of ownership that can arise around ML models: first party ownership, in which the model is internally developed; second party ownership, in which you own some code but are working with another entity that owns the IT model; and third party, in which you are entirely dependent upon another entity that builds the models. 

From there, Shahi laid out use cases for each of these ownership models at Atlassian.

For a first party example, Shahi pointed to a natural language process (NLP) model, which focuses on teaching computing technology and AI to understand text in the same way that humans do. At Atlassian, Shahi utilized this NLP model for community-focused sentiment analysis and topic modelling, drawing on input from across the workforce. He also used it to help fine-tune support processes through text classification, ensuring that tickets enter the right queues for speedy resolution. 

Next, Shahi pointed to Atlassian’s use of Optical Character Recognition and Intelligent Character Recognition (OCR/ICR) software to help automate the processing and archiving of unique handwriting on paper documents including contracts, finances, acquisitions, and beyond, reducing tedious human involvement in completing such tasks.

Lastly, Shahi pointed to the creation of chatbots to help provide around-the-clock support for digital workers across IT, HR, and beyond. To get more in-depth knowledge about these ML use cases and more, you can listen to the full session here.

Justin Kamp
About Justin Kamp

Justin is a freelance writer who primarily covers the intersection of art and technology.