A large energy multinational boosted procurement efficiency by 40% using a GenAI-powered Senseforth solution

Summary

Our client, a large energy multinational, faced significant challenges with outdated and siloed systems that were not designed for rapid information retrieval. This hindered the procurement team’s productivity and caused a ripple effect across the entire organization, impacting operations that rely on the timely and cost-effective sourcing of goods and services. Procurement, the lifeblood of the energy sector, requires efficiency and agility to fuel organizational success. 

To address these issues, Fractal introduced Senseforth. This AI-powered solution leverages the power of large language models (LLMs) and generative AI to revolutionize knowledge retrieval within large enterprises.  

The implementation of the Senseforth solution has led to a remarkable 40% improvement in information retrieval efficiency, streamlining processes, and enhancing decision-making capabilities. This strategic integration of advanced AI technology has not only optimized the procurement function but also propelled the organization towards a more agile and informed operational framework. 

Challenges

The client’s procurement team grappled with several challenges that significantly hampered their efficiency.

These challenges included:

A large energy multinational boosted procurement efficiency by 40% using a GenAI-powered Senseforth solution

Manual knowledge retrieval

Clarifying procurement policies and locating supplier information was a cumbersome process. Staff were flooded with requests and forced to conduct manual searches through an extended network of disparate documents and databases. This time-consuming process diverted valuable resources away from core procurement activities.

A large energy multinational boosted procurement efficiency by 40% using a GenAI-powered Senseforth solution

Operational delays

The inefficiencies in information retrieval led to operational delays. Additionally, the excessive time spent searching for information resulted in bottlenecks within the procurement cycle, which impacted the business’s agility and responsiveness.

A large energy multinational boosted procurement efficiency by 40% using a GenAI-powered Senseforth solution

Decision-making roadblocks

Critical procurement decisions were often postponed due to a lack of timely and accurate information. This led to potential missed opportunities, delays in securing essential goods and services, and financial setbacks.

Solution

To solve these problems, Fractal implemented its Senseforth solution. This AI-powered solution is designed to revolutionize knowledge retrieval within large enterprises using the power of large language models and generative AI. 

In eight weeks, the Senseforth solution was deployed and customized on the client’s AWS environment, where it was integrated with the internal SharePoint and policy repositories and several other backend tools. The virtual assistant was further customized by training it on thousands of documents covering various topics such as procurement policies and supplier agreements. 

Fractal built an advanced RAG solution that leveraged semantic search based on OpenSearch and Bedrock Hosted Models like llama3 and Mistral, ensuring that the answers were relevant and consistent with the source documents. This approach ensured that no confidential or company-specific information left the company tenant. 

Senseforth, built on Amazon EKS, Amazon Bedrock, and Amazon Open Search Service, ensures efficient and accurate information retrieval without extensive training. It seamlessly handles complex documents, including those containing tables, across multiple file formats such as PDFs, Docs, and PPTs.  

Results

  • Rolled out to 1500 users
  • 40% improvement in information retrieval efficiency
  • 20% overall compliance enhancement with a reduction in errors. 
  • 30% efficiency boost in the procurement process
  • 10% increase in overall procurement team satisfaction
Telecom company

Business problem and objectives  

Our client, a large telecom company, needed to implement a Responsible AI framework for two specific use cases: Authorization to Operate (ATO) and Fraud Detection. 

The primary objective was to develop a comprehensive framework that outlined Responsible AI best practices and standards from ideation to decommissioning. Additionally, the client wanted to implement a Governance Framework that assessed scalability and integrated platform and custom changes for both legacy and new solutions. 

Key challenges

The client faced several challenges in implementing Responsible AI, including ensuring adherence to correct processes, aligning diverse teams and stakeholders, and establishing governance-compliant procedures. These tasks required attention to detail, thorough documentation, and leveraging platform capabilities for ethical deployment and management in production environments.  

The complex nature of these challenges emphasized the need for a strategic approach to address the inherent complexities of Responsible AI implementation. 

Solution

To tackle the challenges, we conducted a thorough analysis of the existing platform and identified areas related to governance, metadata management, model registry, and logging. Based on this assessment, we designed a recommended architecture that was not tied to any specific platform, thus reducing the risk of future challenges associated with platform changes or upgrades.  

We developed Standard Operating Procedures (SOPs) to cover various aspects, including monitoring, scalability, ML development, and reusability. Additionally, we operationalized and automated the governance framework in collaboration with our client’s operations team, ensuring seamless integration with existing processes and workflows. 

Impact created

Introducing a Responsible AI framework, checklists/scorecards, and SOPs allowed the client delivery team to swiftly enhance their Responsible AI practices.  

Streamlining the onboarding process for new projects resulted in a 45% reduction in assessment effort by establishing entry and exit criteria.   

Additionally, setting up a Responsible AI Center of Excellence (CoE) led to a 40% boost in use case efficiency by adopting industrialization techniques at scale. 

“Forecast Accuracy

Background

A U.S.-based CPG company wanted a solution that could deliver high forecasting accuracy across multiple categories.

Approach

Fractal used Foresient, its AI forecasting and planning platform with extensive use of AWS services.

Solution Framework

EC2 provided scalable compute capacity for hosting the front end, back end and API’s of the application. EMR based on Spark was used to process and analyze huge volumes of data. The RDS service provided a cost-effective database and S3 stored information in the form of forecasts and models.

Outcome

Foresient generated weekly forecasts by SKU, customer, channel and region-at-scale, which delivered actionable client insights into short, mid and long-term scenarios. Typical forecast accuracy is between 60% and 80% within CPG categories. Foresient on AWS delivered accuracy of 85% to 95% across six different categories.

Test & Learn

Background

A leading U.S.-based retailer wanted to test the impact on sales of removing or lowering a 25% discount on electronic accessories.

Approach

Fractal used Trial Run – a cloud-based testing solution – running on AWS.  Data feeds were established to experiment with various scenarios using POS, store master and product hierarchy data.

Solution Framework

  • The solution, running on AWS, utilized components including EMR, S3, Aurora, and PostgreSQL.
  • Fractal tested 44 locations in all – 22 stores with a lowered discount and 22 stores with no discount.
  • Trial Run’s synthetic control algorithm created the best matching control store against each test store. Control matches on 26 weeks pre-period based on socio-economic score, household median income along with all category combinations.

Outcome

The trial showed negative business impact when reducing the offer from 25% to 15%, as there was a statistically significant drop in both margin and unit sales. However, the 22 stores with no discount during the eight-week trial period, showed a combined increase in margin of $26,000. Expanded chain-wide, this delivered $2.6 million in additional profit.

Speed for desicion

Background

A client needed a faster, self-sufficient way to analyze and understand the performance of its topline KPI’s, split across zones, countries and vessels.

Approach

AWS was selected as the cloud platform to run Fractal’s Cuddle.ai software solution because of its end-to-end availability of key infrastructure components required to deliver speed and high performance. Cuddle.ai was deployed to help the client access information faster, reducing time-to-insight from weeks to days.

Solution Framework

Some of the key components in the solution included EKS – to automate deployment, scaling and management of Dockerized Microservices; EMR and S3 – for data storage and high-speed processing; and RDS Postgres – to store semantic knowledge of the key business concepts used by the client.

Outcome

The client increased speed of decision-making for key executives without relying on a dedicated internal analytics team. Cuddle’s natural language search helped executives query their data easily, using natural language, while sending real-time alerts on unexpected changes to KPI’s.

Forecasting at speed and scale for NielsenIQ
banner letter Contact Us

Introduction

Leveraging advanced forecasting for market adaptability

With more data available than ever before, enterprises must leverage past insights into strategic future plans to stay competitive. As society and consumer behavior evolve, businesses need sophisticated forecasting platforms to adapt and meet market demands. Flexible solutions are essential to handle new, varied challenges and provide insights for future actions. Thus, while most businesses track performance, the key to business success in a rapidly evolving market is harnessing the available data through a sophisticated forecasting platform and plan for continuous improvement.

The challenge

Achieving higher forecasting accuracy

NielsenIQ was looking for a solution that would enable higher category forecasting accuracy, provide ongoing reporting to track performance, and enable strategic planning across their platforms. NielsenIQ specializes in global measurement and data analytics and is the definitive source for retail and consumer intelligence.

Maintaining competitive edge

As an industry leader, NielsenIQ is always striving for an extra edge to drive their business forward, stay ahead of the competition, and continue to be the trusted adviser to the world’s leading consumer goods companies and retailers in an ever-changing business landscape.

Integrating strategic planning

Our experience in engineering bespoke solutions through our integrated cross-market AI-based suite of algorithms meant we were the ideal partner to stand alongside NielsenIQ, and help them approach their decision making with confidence that they could pass on to their own clients.

Solution

Fractal’s Foresient: AI forecasting and planning platform

Fractal’s Foresient, an AI forecasting and planning platform on AWS, generates forecasting models and forecasts at speed with high accuracies. Armed with this solution, we began by looking at historical reporting which allowed us to learn more about NielsenIQ’s categorisations, and the way that NielsenIQ works with its clients to provide the data they need to make actionable business decisions.

Understanding market disruptions

Alongside this, we considered other sets of associated data that could also affect how consumers reach their purchase decision, including short, medium, and long-term impacts that can cause market disruption, for example, Promotions; Prices; World events; Even the weather!

Our automatic data pre-processing and feature selection can be significantly more accurate than simple or manual forecasting, giving NielsenIQ the confidence to forecast for any eventuality.

We approached the challenge by:

Step
  • Understanding the problem
  • Data preparation
  • Forecasting Iterations
Description
  • Problem statement
  • Performed sanity checks, completed missing data, and harmonized data.
  • Created multiple iterations using the AI/ML forecasting engine.
Outcome
  • High-quality, consistent data.
swiper next
swiper prev

Optimizing forecasting operations with scalable infrastructure: To ensure robust and scalable infrastructure for our application for end-to-end forecasting operations and to help businesses focus on ROI, we leveraged several key services.

Component Description The Details
EMR based on Spark

Used to process and analyze huge volumes of data

Efficient data processing

Amazon Relational Database Service

Provided a cost-effective database

Cost efficiency

S3

Stored information in the form of forecasts and models

Reliable data storage

The results

The immediate impact

Higher forecasting efficiency for actionable insights

Foresient generated forecasts for five categories across the United Kingdom and three categories across Ireland, with enhanced accuracy, which delivered actionable insights.

The long-term impact

Flexible forecasting and integration with Foresient

NielsenIQ could access visual forecasts on the Foresient user interface, use the integrated API, or download the data via CSV to work alongside their current workflows and business applications. Comparing forecasts with actual values at different hierarchy levels across time ensured that bias and accuracy metrics were considered as part of the overall analysis. The available options demonstrate how Foresient can seamlessly integrate with existing workflows, offering the additional insights needed to differentiate and excel.

CS image