Cognitive Automation Applying Analytics and AI to tasks within the Software Development Lifecycle Abstract The use of purpose-built Artificial Intelligence solutions has already started fulfilling the promise of business process automation across industries. And while many Engineering organizations are busy implementing AIbased solutions for the marketplace, some are beginning to realize the potential that these capabilities have in transforming their own Software Development Lifecycle (SDLC) processes.
Revisions Date Description November 2019 Initial release Acknowledgments This paper was produced by the following members of the Dell EMC Server Validation organization: Authors: Geoff Meyer, Kevin Olalde, Mark Keating, Erik Reyes, and Ramakanth K The information in this publication is provided “as is.” Dell Inc.
Table of Contents Revisions.............................................................................................................................................................................2 Acknowledgments ...............................................................................................................................................................2 Executive summary....................................................................................................................
Executive summary The application of Artificial Intelligence-based solutions has already started to fulfill the promise of business process automation across industry after industry. As Engineering communities begin to realize this potential and implement AI solutions to transform their own internal Software Development Lifecycle (SDLC) processes, they should pay heed to the risks of underperforming against high expectations and over-simplification.
• • • 5 Approach these initiatives with the full understanding that data and existing process is the elemental resource to success. Your data’s cleanliness, maintenance, ongoing capture, and retention is going to necessitate the adoption of enhanced business processes. Plan for useful and valuable improvements along the project path.
1 Problem Statement Businesses are undoubtedly at the cusp of “AI everything”. Industries as diverse as Motorsports and Medicine, have started collecting data for the purpose of analyzing, predicting, and maximizing organizational performance. AI has been silently churning behind the scenes keeping our credit cards and identity protected. AI has moved seamlessly into our homes with the advent of Amazon’s Alexa and Nest.
For AI and Cognitive Automation, however, there’s several new factors at play and all need to be dealt with simultaneously to achieve the dividends of a successful rollout: 7 • Data Science is a new skill set that most likely doesn’t exist in many functional teams. • Data is the elemental resource of AI. The capture, cleanliness, and ongoing health of data demands constant attention and changes in business processes.
2 The SDLC Opportunity Landscape Ben Pring, of Cognizant’s Future of Work Division and author of “What to do When Machines do Everything” likes to frame AI and the future of work in this way: “X + AI”. X is EVERY task that you perform. Your job is to figure out what X is in your organization or project context, then take the lead on assessing the potential improvements to that task that AI can provide.
There is a new class of Test Automation tools for GUI applications that have emerged in recent years that leverage AI techniques such as Computer Vision and Natural Language Processing (NLP). These tools support the testing of mobile and browser-based applications. They offer the promise of increased coverage across a broad range of test categories including: Functional, Performance, Usability, and Workflow Testing.
Figure 4 - Test Configuration Planning Model This planning task is usually undertaken by two to three lead engineers and takes over two weeks to complete. The goal of this planning task is to determine the fewest number of configurations (contain cost), that provide the highest test coverage of the newest features and previously delivered functionality, accounts for most-sold configurations, and factors in the historically challenging configurations.
We’ve started referring to this capability as Precision Testing. And the more data sets that we add to it (astested configurations, change sets, etc.), the more precision it provides in its recommendations.
3 The AI SDLC Playbook This leads us to recommend the following Playbook for organizations to use when navigating the application of AI-inspired solutions to tasks in their SDLC. Using the approach outlined below provides Engineering teams with a means of countering the optimistic and sometimes unrealistic expectations of their leaders and their Subject Matter Experts with a pragmatic, yet goal-oriented approach. Figure 5 - Playbook for AI across the SDLC 3.
becomes especially important if you require the assistance of Data Science skills not currently residing on your team. Why? Because if the data science skills are coming to you via Consulting Services providers, there’s a natural inclination for them to recommend solutions that they’ve previously implemented, which may or may not be top priorities for your organization.
3.3 Enabling In a typical Software project, this phase would be called the Develop phase. However, Cognitive Automation initiatives carry additional activities that need to be done in coordination. We call this the Enabling phase. Figure 8 - Enabling Phase Monitoring and calibrating the efforts of your project’s efforts using the guideposts provided by the pre-defined Value and Business Case is done throughout this phase.
3.4 Realizing Once the Enabling phase has completed with the development of the algorithms, the establishment of your data sources or data marts, revamping your business processes, and automating the data feed, you are ready to begin the adoption rollout. If you’ve nurtured deep stakeholder engagement throughout, this is where your investment in that engagement will start to pay off. Your early stakeholders are your trusted champions and should be at the forefront of extolling the benefits of adoption.
4 Summary The most important thing to do when embarking on an AI-inspired, Cognitive Automation initiative is to Start with Why. Pinpointing your pain points that provide the most value to your organization also provides the foundation and North Star to establish deep stakeholder engagement. Capture, Manage and Retain your data. Start capturing data now and be thoughtful of the data that you retire or eliminate. Pick your Partner.
5 List of figures Figure 1- Evolution of Automation in the SDLC ...................................................................................................................... 6 Figure 2 - Assessing SDLC Opportunities for AI..................................................................................................................... 8 Figure 3 - AI-Assisted UI Automation ..........................................................................................................................
6 Recommended Reading 6.1 Books • • • • • • • 6.2 • • • • • • • • • • • • • 6.3 • • • • 18 Predictive Analytics: The Power to Predict Who Will Click, Buy, Lie, or Die: https://www.amazon.com/dp/B019HR9X4U/ref=dp-kindle-redirect?_encoding=UTF8&btkr=1 What To Do When Machines Do Everything: http://www.whenmachinesdoeverything.com/ Race against the Machine: https://books.google.com/books/about/Race_Against_the_Machine.html?id=IhArMwEACAAJ Super Freakonomics: http://freakonomics.
• • • • • • • • • • • • • • 19 What’s Everybody So Afraid of: http://www.popularmechanics.com/technology/robots/news/a28645/googlesalphabet-astro-teller-ai/ Robots Are Coming for Jobs of as Many as 800 Million Worldwide: https://www.bloomberg.com/news/articles/2017-11-29/robots-are-coming-for-jobs-of-as-many-as-800-millionworldwide The 10 Biggest AI Failures of 2017: https://www.techrepublic.
