Skip to main content

© 2021 Excellerate. All Rights Reserved

The Client

Based in the Midwest, the client offers data analytics and insights solutions based on their proprietary platform. The platform unifies cognitive computing, machine learning, semantic technologies, optimization, and data technologies.

The Challenge

Most business intelligence solutions offer post-action analysis that is of diminishing value in today’s fast-paced markets. The real value of business intelligence solutions lies in predictive, actionable intelligence.

The client needed a platform that would support enterprise executives with data-driven insights fueling their decision-making process. It intends to do this by converging vast silos of data that currently exist within organizations via an engine that delivers optimized solutions using artificial intelligence models.

The client’s cognitive decision-making platform has multi-fold objectives.

It must monitor and analyze current business events and give timely insights and predict events that may affect the business in the near and mid-term. The platform must also identify future opportunities that the company can leverage.

The platform would use prescriptive analytics and alert users of any issues ahead of time. It would also use AI to recognize patterns in enterprise data and spot aberrations and erroneous behavior. These would apply for specific business situations under certain boundary conditions. The solution should also provide a context-aware, personalized solution to resolve any identified issues.

Keeping scalability in mind, the client wished to apply any learnings and solutions that the platform would present across multiple industries and domains.

The Solution

Excellarate’s team of highly trained professionals helped the client setup AWS CodePipeline with its own set of scripts for automated configuration of microservices. This was a one-time activity. In the background, AWS CloudFormation hosted the development and infrastructure configuration. This helped the client drastically reduce configuration time and minimize effort by simply configuring only the parameters they needed for the service and not worrying about how the service is implemented or run.

The team effortlessly fixes pipeline errors and issues using nested triggers. Separate accounts were used to ensure security compliance at the environment level. To ensure control over the process, manual checks are implemented up until the Delivery stage. Post this, the Deployment stage is fully automated up until Production.

The Retail Perspective

The Retail industry is aptly suited for validating the platform.

With wafer-thin margins, large store chains are under immense pressure from continually changing market dynamics. New business models such as manufacturers going directly to consumers are emerging as a successful model.

These players are also leveraging analytics and big data learning to be more customer-centric. The good news for large retail chains is that they are sitting on massive data lakes, which can yield valuable insights that would positively impact their profitability.

The product management team here at Excellarate has substantial experience in engineering solutions for the Retail industry across Fortune 500 clients. This domain knowledge helped the client and Excellarate get on the same page about the goals and the exact solution needed to achieve these objectives.


The MVP includes two AI product offerings – a voice-based conversational interface, and the cognitive decision-making platform. For product validation, Excellarate reached out to industry connects as well as enterprises as per the target market identification. Excellarate assisted the client to onboard one of the largest lingerie retailers in the US as an early adopter. Based on their feedback, Excellarate validated the product concept and further fine-tuned the platform.

The Outcome

To get a better sense of the current technological prowess at the ground level, Excellarate did an exhaustive competitive analysis of the players offering similar solutions. These were mostly enterprise-scale companies that have invested in the uncharted territory of AI, have mature IT processes and the infrastructure to process the enormous amount of data.

IBM seemed to be the only player that satisfied all conditions; however, its Watson engine is generic and not meant for specific business problems. The client saw an opportunity here to set the rules for a nascent but high-potential market.

Excellarate partnered with the client to engineer a solution that can process hot and cold data streams from any client’s data lake. As part of the product-market fitment, Excellarate identified the following use cases in the merchandising and marketing function, on the product, operations, and customer side,

  • Predicting sales
  • Optimizing product assortments
  • Enhancing loyalty programs
  • Streamlining store operations
  • Inventory optimization
  • Personalization using customer segmentation
  • Optimal pricing
  • Cross-sell and up-sell
  • Devising a win back customer strategies, and,
  • Customer lifetime value (CLV) analysis
  • Shaping the product portfolio

Excellarate helped the client create a product concept using a voice-based conversational system that recognizes the human voice and uses NLP (natural language processing) to analyze big data in real-time. The platform looks for trends, operational metrics such as sales, margins, Stock to sales ratios, Sell-through, and GMROII.

Excellarate further extended this feature to include a predictive feature along with a context-aware language interface. For, e.g., If the user asks, “What are the sales of brand X in Q2 of 2018?” The system would not only answer this question but will also present answers to the most likely asked question that might follow – “What were the Gross Margins for brand X?” and “Which store recorded the most sales for brand X?”

The intent here is to engage the user in a guided conversation to analyze the business situation and to offer additional analytical details. Machine learning algorithms would learn the user’s behavior and preferences. The AI system would then learn from the user’s actions to optimize subsequent responses under similar conditions.