Artificial Intelligence

Designing Intelligent Workflows inside IBM Garage

Share this post:

Author: Kim Bartkowski, Design Principal, iX, Global Business Services, IBM A/NZ

In my last post, I introduced a new blueprint framework our teams in the IBM Garage are using to create Cognitive Enterprises using AI. In this article, I will be talking about a new design practice and complimentary artefact to a Cognitive Enterprise Blueprint called Intelligent Workflows. These workflows are how our teams identify opportunities for AI, automation, robotics and IoT to help make business processes more efficient and more useful.

IBM’s Cognitive Enterprise Design Blueprint.
IBM’s Cognitive Enterprise Design Blueprint.

In this post we will focus on micro-moments in a user experience:

  1. To choreograph an enterprise to deliver on the moments that matter for its customers;
  2. And to help identify the right scenario for AI to deliver on its purpose, value and trust to the human it’s serving.

Intelligent Workflow

Whether you are improving a current state or creating a future state, an Intelligent Workflow uses many common design thinking and Lean UX practices. For example, we use Personas and Value Prop canvases to identify people’s needs and the new value we can create by solving those needs. In an Intelligent Workflow, our teams take a Persona canvas and use that to create the role of the digital worker. This ensures that the design of the human/machine relationship is transparent to the team and stakeholders.

What is a digital worker?

Digital workers are small machines embedded in digital workflows. We call them assistants because they support and augment the human in charge, by doing small, discreet tasks like automating an email to a supervisor or helping to integrate data from one system to another system. 

These assistants are enabled by AI and are triggered by the sensors on an IoT device. I’ve included an example below of a user story about a digital worker that supports a manager in the area of team rostering for an energy company.

Consider all the impacts when a person calls in sick:

  • The gap in the value chain or workflow from the skills provided by this person
  • The leading indicators of sickness, like the amount of work coming up and the lagging indicators of how many people have been sick in the last week, and
  • The redistribution of work that needs to happen so the team can still deliver to plan with minimal impact to energy production.

This multiple-step analysis and synthesis would take a person, hours to perform and the day would be half over. In this user story, the work is done in minutes.

The use case and user story maps below are some of the artefacts our teams in the IBM Garage use to describe the organisation and patterns of data that need to flow through the AI systems to identify, understand, and apply knowledge and reason to produce an outcome.

Designing the human/machine relationship. Created by Benjamin Ma.
Designing the human/machine relationship. Created by Benjamin Ma.
User story mapping. Created by Benjamin Ma.
User story mapping. Created by Benjamin Ma.

Our product teams of engineers, designers and business product owners work to understand the data that sits underneath the variables to determine the fabric of the intelligence they will be building. They also look at the tasks a human would perform in the scenario and match the intents. Then they can collectively design the AI into the experience, interactions and conversations that will take place in the workflow between the humans and the systems. It may sound complicated but we’ve developed fun tools that help to remove the complexity and fear of doing something that is quite new.

Our Design for AI workshops are made so that all people can participate with all types of skills and backgrounds. We’ve included a set of cognitive playing cards with capability pictures on them that relate to the skills people use to perform common tasks. Often, it’s not one card that is selected from the pile, but a combination of these cards that helps our teams determine how the AI will engage with the human. Once we understand all this, then we can design the features and functions of the intelligence in the workflow.

Cognitive playing cards with relatable human capabilities and tasks.
Cognitive playing cards with relatable human capabilities and tasks.

Often these types of Intelligent Workflows are best communicated and understood by stakeholders through a set of common design artefacts:

  1. A pictorial story or journey that communicates the experience for both the human and the AI. A customer journey map or Golden Thread is often used for this.
  2. A sample set of human and AI user stories that simulate the human/machine interactions and how the relationship will build over time, like the ones shown above.
  3. A Cognitive Enterprise Blueprint that houses the validated moments, data and system required to build each interaction.
  4. Small scale experiments of the micro-moments in the workflow that show what is possible and how the data will be used.

Bringing together enterprise strategy and human-centred design can help you scale and mature your company’s AI transformation. To see how IBM Garage works visit us here.

More stories

Can a bot named Sam help citizen scientists save our seas?

A custom data platform and AI tools help UNEP plumb the depths of global water pollution By Jennifer C. Clemente With his furrowed brow and direct gaze, Sam pauses thoughtfully before answering a seemingly innocent question: “Why do you care about marine litter?” “Good question. I care because the people who created me, the men and […]

Continue reading

Designing Intelligent Workflows inside IBM Garage

Author: Kim Bartkowski, Design Principal, iX, Global Business Services, IBM A/NZ In my last post, I introduced a new blueprint framework our teams in the IBM Garage are using to create Cognitive Enterprises using AI. In this article, I will be talking about a new design practice and complimentary artefact to a Cognitive Enterprise Blueprint […]

Continue reading

Diversity needed to tackle the inherent bias in artificial intelligence

AI is designed to make our lives easier. But there’s a big problem when the people who design the systems program in their own bias. Social distancing has been one of the most significant themes of 2020, and it’s not limited to human contact. In June, three technology giants distanced themselves from the multibillion-dollar business […]

Continue reading