Put AI to Work
April 15-18, 2019
New York, NY

Toward automated AI planning in enterprise: Opportunities and challenges

Maja Vukovic (IBM)
4:55pm5:35pm Thursday, April 18, 2019
Machine Learning, Models and Methods
Location: Grand Ballroom West
Secondary topics:  AI in the Enterprise, Models and Methods

Who is this presentation for?

  • Solution architects, technical executives, and business managers planning to implement AI planning



Prerequisite knowledge

  • Experience using AI to develop automation solutions

What you'll learn

  • Understand the basics of AI planning
  • Explore opportunities for AI planning and innovations in this space
  • Learn why using AI planning can improve automation
  • Discover barriers to adoption of AI planning (e.g., world model creation)
  • See how to structure teams to successfully implement AI planning solutions


Existing AI-driven automation solutions in enterprises employ ML, NLP, and chatbots, but AI planning offers an opportunity to drive reasoning about action trajectories to help build automation. AI planning is a problem-solving technique in which knowledge about available actions and their consequences is used to identify a sequence of actions, which, when applied in a given initial state, satisfy a desired goal. There are three main inputs to a planner: initial state, goal condition, and domain description. AI planning has success stories in a number of domains, ranging from space applications, logistics and transportation, and manufacturing to robotics, scheduling, e-learning, and service composition.

The common application of AI planning is for environments where there are many actions to choose from and many ways to achieve a goal by taking a series of those actions. These actions are fine-grained and can be taken in many different situations or contexts throughout the process, and it isn’t clear which action to take in each situation. In this case, you need to understand how these actions may take you closer to the goal. Similarly, AI planning is also applicable in situations where you are faced with complex goals that need to be broken down into multiple ones.

Maja Vukovic demos an application of AI planning for the migration of legacy infrastructure to the cloud, based on real-world examples and data, and discusses challenges in adopting AI planning solutions in the enterprise. The complexity of migration tasks makes this transformation long and challenging, as the environmental conditions continuously change (the availability of landing pad, application release cycles, etc.). Often delivery project managers continuously and manually replan the migration project activities, which is tedious and time-consuming when considering the large scale of transformation projects.

Join Maja to learn how AI planning can expedite and simplify the migration planning process by defining the clients’ constraints and resources in a simplified format that abstracts the user’s need to hardcode domains and problems. This capability is exposed as a service and evaluated for migration plans for multiple clients with varying constraints in the span of a few minutes, thereby enabling migration project manager and migration architects to reason about potential migration plans and automatically replan as needed.

Photo of Maja Vukovic

Maja Vukovic


Maja Vukovic is a research manager and a research staff member at the IBM T. J. Watson Research Center. Maja’s research expertise is in IT service innovation, AI planning, crowdsourcing technologies, API ecosystems innovation, and social media applications for disaster management. Maja leads the cognitive service management team, focusing on AI-driven insights and automation in hybrid cloud systems. Maja has received numerous IBM Outstanding Technical Achievement Awards and IBM Research awards for her technical leadership. She’s a member of the IBM Academy of Technology and a senior member of the IEEE. In 2018, she was recognized with the IEEE TCSVC Award for Women in Services Computing. She’s an IBM Master Inventor, with over 160 patents filed and 50 granted. She’s also the author of over 90 papers in top international conferences and journals. Maja is a cofounder of a number of workshops, including Enterprise Crowdsourcing and Ubiquitous Crowdsourcing and Social Web for Disaster Management, collocated with leading international conferences. Previously, Maja was a research scientist at the Mercedes-Benz Research and Technology Center in Palo Alto, working in the field of telematic services. Maja holds a PhD from the University of Cambridge, UK, for her work on context aware service composition using AI planning, as well as an MSc from International University in Germany and a BSc from the University of Auckland, New Zealand.