Toggle Menu

Rapid AnalytiX Framework


Learning is the Essence of Organizational Agility

The pace of change is accelerating. Data is getting bigger. Networks are getting denser. Information abounds. But to use it advantageously, you must learn how to make it fit your organizational needs. And you need to do it quickly. 

What’s the Problem?

Agile is a proven approach for delivering software solutions. Time-boxed cycles of feedback slice off incremental portions of functionality and rapidly deliver them, making it easier to see progress and validate the solution is developing as intended. Unfortunately, these short cycles can undermine the effectiveness of data science solutions. Data science requires detailed examination of available data, thorough analysis of solution alternatives, and repeated hypothesis testing to determine the best approach. As a result, data efforts tend to emphasize research and learning, work that does not naturally fit into a time-box.

Until now, there has been no effective way to bring these two approaches—Agile and Data—together harmoniously. Excella’s Rapid AnalytiX framework does just that.

 

What is Excella’s Rapid AnalytiX Framework?

Excella’s Rapid AnalytiX Framework is a powerful tool that enables quick and lasting success with AI and Data projects. Rapid AnalytiX captures our extensive experience using Agile principles and practices and Lean Startup techniques to deliver AI solutions. It’s a new way of using these concepts to accelerate learning and harness the power of your data to improve your organization’s performance. 

With Rapid AnalytiX, we work with you and your stakeholders to clearly define the opportunity. We use our rich expertise to examine the potential of your data and explore multiple options in parallel. As we develop potential solutions, we collaboratively assess them using a design canvas and identify the most promising ones. We build those solutions using an automated MLOps infrastructure that allows quick comparisons and validation of the most effective approaches. Once they’re proven, we automatically deploy them into production and monitor their performance to ensure they give you lasting benefit.

Throughout this process, we use our AI Ethics Guidelines to ensure our solutions deliver responsible and ethical outcomes. We review the intended use case, the potential impact on individual and community welfare, and possibilities of bias. When we build our solutions, we ensure they are explainable, transparent, and trustworthy by taking a proactive approach to data, training, monitoring, and privacy. We build in controls to maintain “human in the loop” safeguards for appropriate oversight.

What Benefits does Rapid AnalytiX Offer?

Rapid AnalytiX is a revolutionary approach to delivering AI solutions. It improves the efficiency of AI solution development efforts by coupling the best principles of Agile with the best techniques of data science. In addition, it delivers several specific benefits that improve your organization’s ability to identify opportunities for AI and deliver solutions that meet them.

By deliberately framing the challenge, we bring focus to the “right” problem before building a solution and maximize the value of your investment.

By defining success criteria before solutioning, we make it easier to pivot to more promising alternatives and accelerate the delivery of value.

By using a design canvas, we enable a systematic comparison of solution options and design alternatives and select the best for you.

By using our “zero baseline” (should consider trademarking this term) approach, we get quantitative feedback on each model iteration and determine whether to persevere or pivot.

By coupling two loops together, we allow for iterative (Iteration) and incremental (New Functions and New Data Release) improvement of your AI solutions.

By using MLOps, we automate processes for training, evaluating, and deploying AI solutions and ultimately accelerate their development.

By using MLOps, we ensure the health of AI solutions by monitoring their performance and triggering automatic retraining when drift or other undesirable outcomes are detected.

By employing our AI Ethics Guidelines, we deliver AI solutions with responsible and ethical outcomes that are explainable, transparent, and trustworthy. We’ve written more about the importance of Explainable AI (XAI) here.

How does Rapid AnalytiX Work?

Rapid AnalytiX solves the problem of getting Agile and Data to work together, enhancing the effectiveness of both. Rapid AnalytiX is a powerful combination of Agile Delivery methods and Lean Startup techniques. It integrates the best features of both into a series of learning loops. These loops rapidly explore alternatives to identify the best opportunities and capitalize on them, so we use Agile’s emphasis on iteration and learning to enhance the research and analysis of data science. That allows us to quickly deliver solutions that balance desirability, feasibility, and viability. 

The Rapid AnalytiX framework delivers valuable solutions through seven steps:

Excella red number 1 icon

Ideate

In this optional stage, we work with your stakeholders to identify your most pressing needs and the best solution concepts for addressing them.

Excella red number 2 icon

Explore

Once we have identified a promising opportunity, we explore your environment, your data sources, and your infrastructure to develop the best solution alternatives.

Excella red number 3 icon

Solution

We capture the most promising solution ideas, including their ethical implications, on design canvases. In collaboration with you, we prioritize them and decide what to build.

Excella red number 4 icon

Build

We build the best solution approaches in parallel, using timebounded iterations and automated MLOps infrastructure to accelerate progress.

Excella red number 5 icon

Validate

At the end of each iteration, we assess what has been built and determine whether to continue improving the solution, pivot to an alternative, or move to production.

Excella red number 6 icon

Deploy

When a solution approach is proven, we use automated MLOps infrastructure to deploy it to production.

Excella red number 7 icon

Monitoring

We use MLOps infrastructure to proactively monitor the performance of production models and ensure they continue to deliver desired results.

Each of these stages is coupled together through a series of Rapid AnalytiX Learning Loops:

Idea Refinement: We cycle between ideation and exploration using your data as a guide to identify the most valuable problem to address and refine potential solution concepts.

Iteration: Our iteration loop is a timebounded build and validation loop. We use it to make our progress visible on a regular cadence and ensure we are moving in the right direction.

Pivot: If our solution approach is not achieving the desired goal, we discard it and pivot to alternatives.

New Functions and New Data Release:  We will continue to refine and enhance production models with additional features and data sources while it is valuable to do so.

Retraining: Production models may drift away from desired performance because of new data or new circumstances. Our monitoring will automatically detect this drift and trigger retraining to correct it.

Contact Us

Contact us to learn more about how Rapid AnalytiX can help you identify and solve your most pressing AI and Data needs.

Christina Seiden
VP, Strategic Growth
Claire Walsh
VP, Engineering and Services
Mathias Eifert
Managing Consultant and Technical Fellow