The IBM Garage expands to Data Science Insights

Share this post:

In the IBM Bluemix Garage, many things change day to day: the companies that come into our Garage locations, the innovative ideas that we are working on, developing, and testing, and so much more that goes on behind the scenes. As Savannah, one of our veteran developers in our San Francisco Garage once wrote, “To stay relevant, businesses,” including ours, “must learn how to keep up with the rapid growth of technology.” The Garages started with building apps on Bluemix, and since the beginning we have been expanding our scope, skills, and pulling in expertise from across IBM. We have expanded to focus on IoT, cognitive, and APIs. Our most well-known expansion is our extensive work on blockchain. Today I am pleased to announce that we are expanding and formalizing our focus on data science insights and analytics.

In the Garage, we focus on proving new ideas quickly – both technical proof and business value. At the heart of what we do is applying IBM Design Thinking to determine a hypothesis to test and Minimum Viable Products (MVPs) to build. We then apply leading DevOps and agile practices within an interdisciplinary squad to deliver code to test said hypothesis rapidly. This approach is hardened and published in the Cloud Garage Method. In partnership with IBM Analytics Services, we have applied our method to create a new offering that helps clients turn data into actionable insights quickly. The Garage’s Data Science Insights MVP supports clients in mining value and insights from their troves of data to make better-informed decisions, address challenges, and uncover new opportunities. And, in the process, we leverage best practices, automation, and analytics patterns to deliver a secure cloud environment for data discovery that clients can use well after the engagement is complete.

Applying the Garage Method to develop meaningful data insights

Just as we did with blockchain, we modified our core Design Thinking and Architecture workshops, developed architectures and data analysis patterns, created enablement, and developed roadmaps for production deployment unique to data science. Our worldwide Garages are already helping clients identify actionable business insights and build analytics and data science models that prove they can be obtained from the available data sources – including a workshop in our London Garage this week! And we’re not done, additional data science and analytics content will be published in the Cloud Garage Method soon.

In developing this offering, we partnered with clients to measure the impact that having faster insight to their data would provide. We worked with a municipal government to improve their revenue tracking, helping them decrease inefficiency and increase overall revenue by analyzing non-traditional sources of data to determine if revenue was being properly reported. We next collaborated with an industrial supply delivery company to prioritize information displays and allow its team to focus on the right priorities.

The Garage Data Science Insights MVP journey starts by identifying the use case, problem, or opportunity you want to focus on. Through a series of discussions, a visit to one of our nine worldwide Garages, or an IBM DataFirst Discovery Workshop at your location, we work with you to determine the high-level scope and data availability, access, and security that can be applied to the desired use case.

Once the scope and data information have been defined, we execute a three-and-a-half-day Garage Data Science MVP Workshop, ideally with your business experts, IT architects, and data experts in the same room. We apply IBM Design Thinking to create a user-centric discovery process, and identify the hypotheses to test and define the ‘minimum viable insights’ needed to demonstrate value to the target users. The data sets are often new to an organization – client systems, licensed data such as our own Weather Company, or public resources – and are combined with existing data. We then dive into the technical architecture: what data is needed to derive insight, where the data is located and how it can be accessed, what security is needed for each data set, which selection of the model and data services, and where the initial solution will be deployed – private, hybrid, or public cloud.

Then during the Insights MVP Build-Up the excitement of immersing yourself in data begins! Over six weeks, the project team will build and refine the identified data science models, and rapidly provision the environment to ingest data to gain insight. The environment will use IBM cloud data services, most often including the IBM Data Science Experience. And of course, we ingest the data needed for the models with all the required security. An expert from your team, who is knowledgeable on the domain and desired insights, works closely with our technical team and continuously provides validation and direction. They are ideally co-located for at least part of the time, or are in close touch by video conferencing. Our data scientists and data services experts can do all the work of setting up the environment and developing/proving out the models. However, there is additional value by having your data team collaborate with our experts, learning directly from our practitioners. We also make enablement available to our clients on data science and AI as a complement to this offering through our online Cognitive Curriculum course, in partnership with Galvanize. At the end of the Insights MVP Build-Up, you will have validated the model and discovered how insights produced could help consumers of the data make more informed and / or predictive decisions.

Having validated the insights’ value, the next phase is to operationalize the model and deliver it to the target end users on an ongoing basis getting to a full production implementation. The Garage will develop the production architecture, plan the setup of the production environment, design how the insights will be delivered to the end users, and provide input to your business case. Getting to production is different for every solution, ranging from getting insights from Twitter data or product information (quick and easy), to insights on consumer buying behavior or leveraging sensitive data (slower and complex).

Sometimes Insights MVP Build-Ups will result in the inability to develop the desired insight, or, as you test with the target end users, the hypotheses about the insight’s value is proven to be incorrect. You can use these results to decide on your next steps – maybe you pivot and look for other insights; maybe try a different model; or maybe you revisit what the end users need.

After seven weeks, you can proceed to your next step with confidence on technical viability and business value. The Garage Method drives alignment across your business and IT organizations so everyone will have been on the seven-week journey together, seen the results, and can rapidly move on next steps. Additionally, your team will have gained expertise on the Garage Method for rapid innovation and on data science for informed decision making.

Where to go from here – for us and you

As I wrote previously about the Garage: “This is how we grow at the Garage – we listen to client needs; we innovate and experiment with experts leading the way; we validate and refine; and we codify what we learn and bring it to clients digitally and through consulting.” Personally I’ve had a great time working with our data experts and embedding their expertise into the Garage Method. In a few months I hope to share a client reference video and blog with you on the Data Science Insights MVP that is as positive as Bendigo and Adelaide Bank’s experience with the Garage and American Airlines use of the Garage Method.

Our Garages are located in dynamic innovation communities in cities around the world: San Francisco, Austin, New York City, Toronto, London, Nice, Singapore, Tokyo, and Melbourne. Scheduling a visit to the Garage is easy and free: come learn how we can help you enact data-driven innovation. You can learn more about the Garage and get in touch with us through our website or via email with our team directly at

More What's New Stories

Compose for JanusGraph arrives on Bluemix

Bluemix is home to some of the most powerful databases in the world. Today, we are expanding that range of databases and are proud to announce that JanusGraph is arriving on Bluemix. It is coming as one of the Compose databases, bringing with it the power of fully open source graph databases.

Continue reading

Vertical Scaling control for Compose databases on Bluemix

As we've built Bluemix Compose, we've been extending the custom Bluemix dashboards for each of the Compose databases. We've been addressing the most requested features from our users. We are, therefore, pleased to announce that one of the most requested features — vertical scaling control — is now available.

Continue reading