This session introduces AI-driven automation and looks at the building blocks needed to automate operational tasks and decisions.

It discusses ground-breaking innovation that opens up the next stage in data and analytics for the data driven enterprise. It provides key information on how to use data, analytics and AI to automate decisions to significantly shorten time to reduce costs, reduce risks and seize opportunities to grow revenue.

Target Audience: Chief Analytics Officer, Chief Data Officer, Data Architects, Data Scientists, Business Analysts
Prerequisites: Basic knowledge of analytics and machine learning
Level: Advanced

Extended Abstract:
According to some analysts, the hyper-automation software market opportunity in 2025 will be as much as $850Bn. There is no doubt that if you can scale automation it will help companies significantly reduce costs and outperform their competitors in revenue growth. There are many things that can be automated including tasks, processes, IT and DevOps operations. However, there are so many things needed to make this happen. Given this opportunity, this session introduces AI-driven automation, looks at the building blocks needed to make it happen.  

• What is AI-driven automation and what can it do?  
• Size of the AI marketplace
• Use case opportunities
• Automating human tasks using AI-driven robotic process automation
• Intelligent document processing
• The power of human and AI-driven orchestration
• Automation building blocks
• Business goals
• Process mining  
• Data Events, business conditions and event detection
• Data integration
• Machine learning model services to predict outcomes
• Decision services
• Action services (skills that can be automated)
• Intelligent Orchestration - goal driven closed loop automation  
• The AI-driven automation process - capture, prepare, analyse, decide, act, optimise
• Human and AI-driven decision and action automation
• Getting started - what needs to be considered?

Got great ML, analytics, and engineering talent, but need to increase the adoption of the ML and analytics solutions your team produces? Wondering how to design decision support applications and data products that actually get used and generate business value?  If you're tired of making 'technically right, effectively wrong' data products that don't get used, this session will help! 

Before you can generate business value, your data product first has to be used and adopted. That success boils down in part to the UX you afford users. After all, the UX is the perceived reality of your data product. However, the skills for designing a great UI/UX are different than those required to do the technical side of analytics, AI/ML, and engineering. Users don't want your data outputs; they want clear answers and actionable decision support—and that’s what we’ll learn how to do together. 

The workshop is a reduction of my full 8 week training seminar. In the ½ day workshop, we will focus on learning 3 main skills via a mixture of lecture, peer discussion, and active exercises/participation. You will learn:

• How to measure your data product’s utility and usability so that everyone on the team has a shared understanding of what a “good UX” is and how it will lead to business value 
• How to use 1x1 interview research to uncover hidden stakeholder and user needs before it’s too late (and your solution can’t be easily changed)
• How to use low-fidelity prototyping and sketching as a means to get aligned with your users and stakeholders and avoid building an incorrect “requirements-driven” solution 

MAXIMUM ATTENDEES: 48

MATERIALS YOU WILL NEED:

• A laptop is required for participation
• Willingness to participate in activities that require pair learning
• Willingness to be open and share with your table and the room when called upon to contribute
• For best results, you should have some sort of strategic decision support application, data tool, or data product in mind to which you hope this training can be applied when you return to work. Design is best learned through doing, and having a real project to apply it to will accelerate that learning. 

Target Audience: Directors, VPs, and principal data product leaders building custom enterprise data products and decision support applications for which adoption is critical to success and the generation of business value. Participants often come from ML and software engineering, analytics, and data science domains, yet also have a responsibility to ensure solutions are useful, usable, and valuable to the business. The training will not help you if you're interested in only working alone on implementation, coding, statistics, modeling and making outputs without any regard for whether they serve the audience they are intended to help.

Prerequisites: You're ready to approach your data work differently, with a human-first, data-second approach. You don't think that the reason that data tools/apps/dashboards go unused is because the users aren't 'smart enough' to understand them. You believe it's more interesting, fun, and valuable to produce data products that actually get used, produce value, and change people's lives. You're curious and open to non-analytical approaches to problem solving.

Level: Expert (you can be a design novice but should be a leader in your core field)

Extended Abstract:

Want to increase the adoption of your enterprise data products? 

It's simple: your team's AI/ML applications, dashboards, and other data products will be meaningless if the humans in the loop cannot or will not use them.

Yes, they may have asked your team for those ML models or dashboards.  

Unfortunately, giving stakeholders what they asked for doesn't always result in meaningful engagement with AI and analytics -- and data products cannot produce value until the first hurdle is crossed: engagement.  

Until users actually use, trust, and believe your ML and analytics solutions, they won't produce value.  

'Just give me the CSV/excel export.' How many times have you heard that -- even after you thought your team gave them the exact ML model, dashboard, or application they asked for?
No customers want a technically right, effectively wrong data product from your team, but this is what many data science and analytics teams are routinely producing -- because the focus is on producing outputs instead of outcomes. The thing is, the technical outputs are often only about 30% of the solution; the other 70% of the work is what is incorrectly framed as 'change management' or 'operationalization' -- and it all presumes that the real end-user needs have actually been surfaced up front.  

If you want to move your team from 'cost center' to 'innovation partner,' your team will need to adopt a mindset that is relentlessly customer-centered and measures its success based on delivering outcomes. However, this is a different game: it's a human game where ML/AI and analytics is behind the scenes and customers' pains, problems, jobs to be done, and tasks are at the forefront. Enter human-centered design and data product management: the other skills that modern data science and analytics teams will need if they want to become indispensable technology partners to their business counterparts.  

This talk is for data product leaders who have talented technical teams, but struggle to regularly deliver innovative, usable, useful data products that their customers find indispensable.  
You've heard for 20 years how Gartner and other research studies continue to predict limited value creation from enterprise data science and analytics engagements, with 80% of projects on average failing to deliver value.  

MIT Sloan/BCG's 2020 AI research shows that companies who are designing human-centered ML/AI experiences that enable co-learning between technology and people are realizing significant financial benefits. 
 
Leaders aren't repeating yesterday.  

If your data science and analytics requires human interaction before it can deliver any business value, you won't want to miss this session with Brian T. O'Neill -- the host of the Experiencing Data podcast and founder of Designing for Analytics.