July 22nd, 2020

In a World Where… AI is an Everyday Part of Business

RSS icon RSS Category: Company, Driverless AI, Machine Learning Interpretability

Imagine a dramatically deep voice-over saying “In a world where…” This phrase from old movie trailers conjures up all sorts of futuristic settings, from an alien “world where the sun burns cold”, a Mad Max “world without gas” to a cyborg “world of the not too distant future”. 

Often the epic science fiction or futuristic stories also have a rich backstory. In the case of successful movies (at least the ones successful at the box office) they also end up generating prequels. And to make a prequel, the writers must figure out what would have had to happen in order to get the protagonists to where and what they are in the future world. 

Consider this: A world where … AI is an everyday part of business. 

That’s the world of the very near future. And that means you are living in the prequel at this moment. What, then, do you need to do now in order to be ready for the AI world that’s just around the corner?

Whatever your expertise – business leader or technical practitioner – there are things you can do now, in this prequel world, to prepare for a world where AI is an everyday part of business

Getting Ready for AI: Where Could It Pay Off?

Everyone’s business situation is a bit different, so the steps you need to take now will vary from what others should do. But there are some basic actions that will help to set you up to take advantage of AI, making sure you are not behind the curve. 

Try this exercise: Be your “future self” for a moment, and figure out what you’d like to have AI to be doing for your business in six months or a year from now. This could be something new, flashy, sophisticated that AI makes possible – maybe a new line of business or a new capability you’ve not attempted before. But much more likely and more practical is to identify something simple that is immediately addressable by AI – a bottleneck in a business process for instance. The question you address and the models you build do NOT need to be complex in order to be valuable. The sophistication comes in knowing your own business well and in recognizing key processes that could benefit from automation, rather than in having to build extremely complicated models. 

A useful target goal might be a process that would be improved by faster decisions made on larger-scale data than humans can reasonably handle. Often the places where AI can most easily pay off is to address basic questions that you already have, as long as you refine them appropriately to be addressable by machine learning systems. And remember: in order for AI to be valuable to meet practical business goals, you must have planned a way to take action based on the insights AI delivers. The action may be human-based or actions that machines take. Either way, action is a key part of a successful design.

One way to come up with good ideas is to look at what others are using AI to do. H2O.ai has been recognized for its completeness of vision and has been named by Gartner as a visionary in the AI space in two of their Magic Quadrants. H2O.ai works with customers across a wide range of industries, with multiple applications of AI even within a single company. Their collection of industry solutions is a rich resource for you to explore to find your own best, first, AI projects. 


The Industry Solutions presented by H2O.ai is organized by industry, by use case (shown here) or by customer story. Visit the solutions resource page to see if you can recognize your own business issues there.

If an AI solution has worked for someone else, there’s a good chance it can pay off for you as well. 

Once you have identified a business goal that is ripe for machine-generated decisions, what’s next?  Ask your future self, “What data do you wish you had started to collect earlier to be ready for training AI models in this future world?”  That tells you what you need to do with data now. 

Collecting Data Now for an AI Future

AI is only as good as the data used to train and run models. 

All too often, people focus on finding the algorithm they want to use and underestimate the importance of – or the effort required to have – good data. You don’t want to have decided on a key question and gotten ready to build an AI system to answer it, only to have your future self realize you’d need to look at a year’s worth of data that you had not thought to collect. Better if you had begun to collect the data in advance – beginning now – if at all possible. 

Of course, you don’t always know what data you will need, but there are some positive steps to take ahead of time that may well set you up for faster AI development when you do begin to build and train models. A wide range of data sources can be useful for AI. You may need data to track customer transactions, the behavior of visitors to your website, IoT sensor or performance, and health of your servers or applications. 

Here are a couple of data-centric things you can do to get ready:

  • Look at the data you already have: Is it what you think? Is it collected together for a comprehensive view? What questions could you ask of that data? A practical approach is to clean up and make use of data you already have.
  • What data is needed to address the questions you choose? Do you already have the data or, if not, where can you get it?  You may need to get permission for access and begin to collect a particular data set or a variety of data from which training features can be extracted. And if you find it’s not practical to get the data you think you’d like, you may need to modify the question you plan to address with AI. 

Building a successful AI system is a balance between identifying the data sources you most desire and the practical considerations of what data is easily available. 

Start the Clock on Building Your AI Experience

Your future self will be pleased to look back and see that you had the foresight to start building experience with AI early on. If you are a technical practitioner — a data engineer, data scientist, or data-scientist-to-be — it’s important to try out different AI or machine learning platforms or tools to gain familiarity and to find the ones that will work the best for you. Even if you have a less technical background, such as a business leader, it’s valuable to set up a test or exploratory project with an organization you may use to outsource your AI project. The sooner you start, the sooner you begin to gain the experience you’ll need to get an AI system up and running to address practical business goals. 

Can you afford to experiment?  Yes! It is essential to make room for experimentation if you want to take advantage of innovative approaches such as AI. But what if the initial experiment with AI does not produce a viable product? You must find a way to make that an acceptable outcome. Remember, a “no failure policy” is also a  “no innovation policy”. 

How, then, do you make it possible to experiment? Part of the answer lies in re-defining what you mean by a “deliverable”. Whether your team experiments with AI/ML technologies themselves or try working with outside experts, your deliverable does not have to a a product or in-production system in order to be valuable. The deliverable can be framed as specific, communicable experience.  

This does not mean trying anything that comes to mind, willy-nilly, with no limitations on time or costs and no specific goals.  Like any engineering, experimentation with AI is and should be limited by available resources, time constraints and budget. But the trick is to think of the acquisition of experience as a concrete and valuable thing.  To do that, you must have a way to communicate what is learned, a way for the experience to ripple through your organization. Make a plan for how you will apply the results of the experience – whether positive outcomes or better understanding of what causes failures –  to the next project which may (should) be an actual AI system built for production.

How Can We Help?

Here at H2O.ai, we work with people in all stages of the process of transforming their company into a data-driven, AI-rich organization.  You can learn how customers such as Wells Fargo, Kaiser Permanente, PayPal, ING or Stanley Black and Decker are using AI by checking out customer testimonials or use case briefs. You also can reach out to talk with our team about your business and how AI may be put to work to your advantage. 

If you have data scientists in your organization, invite them to try the open-source AI platform known as H2O-3.  Experienced and novice data scientists will find it valuable to get their hands on the H2O Driverless AI platform for a free tutorial or a 21-day free trial . Driverless AI lets them quickly experiment and extend their own experience through state-of-the-art accelerated feature selection, model selection and model tuning, coupled with superb AI explainability. There’s no black box here. You can see how models are making decisions, a key aspect of building trust.

H2O Driverless AI makes it easier to develop highly accurate models that are ready to deploy. You can find out more about H2O Driverless AI in this solution brief.

Welcome to building your prequel for a world where … AI is an everyday part of a successful business!

Additional Resources

Read the blog post “Machine Learning in the Age of AutoML” by Parul Pandey 

Download the pdf “8 Key Considerations for AI in the Enterprise”

Register to watch the webinar “What is AutoML?” with Rafael Coss

Read the blog post “AI/ML Platforms: My Fresh Look at H2O.ai Technology” by Ellen Friedman

About the Author

Ellen Friedman, PhD

Ellen is Technical Evangelist at H2O.ai. She is an international speaker, author, and scientist with a PhD in biochemistry from Rice University. Ellen has been a committer for Apache Drill and Apache Mahout projects and previously a laboratory researcher in molecular biology. In addition to authoring publications in technical fields from genetics to oceanography, she is co-author of data-related books published by O’Reilly Media, including AI & Analytics in ProductionMachine Learning Logistics, Streaming ArchitectureIntroduction to Apache Flink and the Practical Machine Learning series. Ellen has been an invited speaker for keynotes at JFokus in Stockholm, Big Data London, the University of Sheffield Methods Institute (UK) and NoSQL Matters in Barcelona as well as invited talks at Nike Tech Talks (Portland OR), Berlin Buzzwords and Strata Data conferences in San Jose CA and London. She's also an artist with not-enough-time for the paint box.

Leave a Reply

5 Key Considerations for Machine Learning in Fair Lending

This month, we hosted a virtual panel with industry leaders and explainable AI experts from

September 21, 2020 - by Benjamin Cox
The Benefits of Budget Allocation with AI-driven Marketing Mix Models

Excerpt of the white paper: “The Latest in AI Technologies Reinvent Media and Marketing Analytics

September 17, 2020 - by Michael Proksch
What it is like to intern at H2O.ai

Blog post by Jasmine Parekh Let’s be honest, 2020 is not going to go down as

September 15, 2020 - by Jo-Fai Chow
My Experience at the World’s Best AI Company

Blog post by Spencer Loggia When H2O announced that remote work would continue through the summer

September 15, 2020 - by Jo-Fai Chow
Desmistificando a Inteligência Artificial e seu papel no sucesso dos negócios

A Inteligência Artificial tem sido um termo bastante utilizado atualmente, mas será que todos sabem,

September 14, 2020 - by Bruna Smith
Modèles NLP avec BERT

H2O Driverless AI 1.9 vient de sortir, et je vous propose une série d'articles sur les

September 2, 2020 - by Badr Chentouf

Join the AI Revolution

Subscribe, read the documentation, download or contact us.

Subscribe to the Newsletter

Start Your 21-Day Free Trial Today

Get It Now
Desktop img