Blog Categories

Bridging the Gap between BI and AI

Gordon Uszkay

VP Data & Analytics

mind the gap image of trains in the uk

Traditionally analytics teams have combined deep knowledge of business objectives and available data to get insights into improving performance. They work with stakeholders by sharing presentations, reports and more recently dashboards, collectively known as business intelligence, or BI.  Together they would create and implement a strategic plan of action around these insights:  

  • adjusting processes 
  • changing marketing approaches to certain customer segments 
  • changing pricing for product lines 
  • other strategic initiatives

The benefits of this data-driven strategic analysis is enormous, but the information and resulting actions require human consumption, communication, and interpretation.  This “human in the middle” approach drives changes at the organization – or “macro” – level, broad strokes that take effect long after the events driving the decisions have occurred.  

AI Fills the Data Gap 

Artificial Intelligence initiatives change the game entirely  Using predictive/prescriptive analytics, data science, and advanced algorithms, analytics teams can design systems to generate very specific insights in or near real-time.  

For example, modeling a customer’s interactions can identify them as a potential growth target or flight risk. An AI system can then immediately respond by contacting customer support or even the client directly!  We experience the incredible power of these automated insights when we shop online at Amazon and other effective consumer facing e-commerce solutions. Analytics teams are now applying this kind of approach to commercial business.   

Bringing these immediate and focused “micro-level” responses requires a whole new set of skills and functions for the analytics team.  

Identifying Gaps in your new AI Team 

Data scientists, including self-taught “citizen” data scientists / business analysts, are a natural extension of the BI team.  They will build and validate the models, likely using a small sample of data and the popular Jupyter notebook software on a personal computer.   

When it is time to implement the new AI platform, however, many other skills will be needed to achieve it’s potential.   

Machine learning (ML)Data and DevOps engineers work together to ensure that the system performs effectively. They will ensure data processing is at the speed and reliability needed.  These functions are typically thought of as part of IT, and the analytics team will certainly be looking for assistance in these areas! 

Here there is another challenge: traditional IT professionals are used to dealing with transaction-based systems and large data sets. In many cases, however, they are not familiar with the very computationally intensive and mathematically complex algorithms that make up an AI solution.  Graphic Processing Units (GPUs), number crunching platforms (e.g. Spark), and ML training platforms (e.g. Airflow) may all be completely new to the IT department. 

They can also struggle with preparing user interfaces and testing plans to properly address the ambiguity and context – confidence, bias, and precision – necessary to interpret the results of AI algorithms. There can be a significant gulf of understanding between the analytics and IT teams, leading to many errors and delays during the development of new AI platforms.   

Better Together: Combining IT and Analytical Skills

While the data scientists can partially bridge the gap, they rarely have the experience needed to architect an enterprise-scale production system – nor the interest!  Filling in this gap requires individuals with experience in high-velocity, high-volume, cloud computing platforms as well as enough mathematical background to efficiently implement data models and validate the results.  This special expertise bridges the gap between the analysts understanding of the models and the IT skills necessary to successfully build and operate AI systems. 

 Establishing and developing a hybrid data science / IT team is an important step in moving an organization to AI maturity. 


The Advanced Analytics Team 

Successfully delivering AI at scale – specifically, analytics integrated into operational systems – requires a special set of skills and experience is paramount.  Indellient established its advanced analytics practice in 2013 and has a multi-disciplinary team consisting of data scientists, ML engineers, front and back end developers, project managers, data engineers, and DevOps specialists who focus on AI projects.  Working closely with our clients, we help bridge the gap between an organization’s analytics and IT teams. This allows them to develop the necessary skills while delivering valuable AI platforms quickly and at scale.   

Join us on September 8-10 2020  for our #DataChampionsOnline series. At home, in the office, on desktop, cell or tablet. 

Insights you won’t find anywhere else, brought directly to you. 

About The Author

Gordon Uszkay is the Vice President of Data Science and General Manager of U.S. Operations at Indellient. Gord leads a team of data scientists, data engineers and developers to deliver insightful and innovative solutions to our customers. Gord has over 20 years’ experience in application development, solution architecture and statistical and predictive analysis. Gord received his BMath in Applied Math and Computer Science from the University of Waterloo and a MSc in Computer Science from McMaster University.