Feature
Companies have long generated and accumulated large amounts of data on their customers, goods and services. But it has not always been possible to make good use of the data, which can be scattered across different platforms or too large. With artificial intelligence (AI), businesses are taking their data farther, using it to improve everything from customer service to logistics and fraud detection. Here are some examples of how AI is transforming various business practices.
1. AT&T
AT&T is revamping its business by incorporating AI internally. The company started offering AI-as-a-service (AIaaS) to internal AI users and has continued to expand it, according to a case study.
The AIaaS platform replaced the open-source software the company previously implemented for its data science team. The previous system relied on a small team of experts and required multiple months to deploy projects that showed a strong return on investment. Since AT&T began using AIaaS, more people within the company have access to AI, and projects based on AI moved from conception to reality faster than before. AT&T worked with H20.ai as the platform for its transformation.
Among the applications the company implemented is a fraud detection solution for iPhones. Some people try to create fake identities to gain access to mobile service and iPhones with little or no payment. AT&T moved from a rules-based system to an AI-based one on H2O.ai’s cloud to create multiple models for detecting fraud with increased accuracy, reducing fraud by over 80%.
“We believe the co-invention and co-development of H2O AI Feature Store is going to be one of the most impactful elements in our platform,” says Prince Paulraj, associate VP of data science, AT&T. “We’re building AI solutions that are much faster, more accurate and robust in a fraction of the time.”
Results
- $7 million annual savings from predictive maintenance on service trucks
- $10 million annual savings from route optimization for service technicians
- Helped Argonne National Laboratory develop flood prediction models without using a supercomputer
2. Migrato
Migrato, a Dutch company that helps organizations process unstructured data to find actionable insights, needed to help its government customers manage new regulations. A new law requires government agencies to publish all the information they hold that might be in the public’s interest — but first redact the personal data in every document, according to a case study.
Migrato already had natural language processing (NLP) classifier software, but it didn’t think the software would scale to handle the volume and complexity of the new data processing workloads the law would demand. The company worked with IBM and its Watson Natural Language Processing Library for Embed, which used natural language AI to improve the performance of existing software, such as Migrato’s. The integration allowed Migrato to clean up customer data to meet the new requirements.
“In the past, our keyword analyses were based on statistical analysis of all the words used in a document, but in many cases, the most common words in a document aren’t the most important ones,” says Oscar Dubbeldam, CEO, Migrato. “We can leverage AI to perform topic keyword extraction, helping us drill down into the meaning of unstructured data.”
Results
- Developed a working prototype in a few hours
- Went live with its solution in five days
- Ready for 400 potential government clients
See more: 5 Generative AI Case Studies
3. Prime Therapeutics
Health care organizations work hard to catch instances of fraud, waste and abuse (FWA) in prescription drugs. Their efforts are hampered by having separate data sources, which can show what drugs a patient receives but cannot put that together with the diagnosis behind the prescription, according to a case study.
Prime Therapeutics wanted a way to improve its FWA detection systems. The company worked with SAS and SAS Detection and Investigation for Health Care, a cloud-based analytics platform. The platform allowed the Prime to apply AI and machine learning (ML) capabilities to disparate data sets and keep problems from falling through the cracks.
Prime identified members who received prescriptions through deception as well as providers who received kickbacks or submitted duplicate claims. The system also made it easier for the company to share evidence of fraud with law enforcement.
“We’ve worked so hard over the years to fight fraud, and this new capability . . . has helped us take it to the next level to safeguard our members and save our clients millions of dollars,” says Jo-Ellen Abou Nader, VP of FWA and supply chain optimization, Prime Therapeutics.
Results
- Over $355 million in recovered payments and cost avoidance for the company’s health plan clients in 18 months
- Identified a person who collected 48 narcotic prescriptions from 10 different providers and 10 pharmacies in one year
- Increased speed and accuracy of investigations
4. Tchibo
Tchibo is a German coffee roaster and seller that also sells an ever-changing array of about 3,000 products, including clothing, furniture, household goods and electronics. The company can encounter logistics challenges in managing supply and demand of the various products it handles, according to a case study.
Tchibo worked with Google Cloud to use its AI capabilities to build an on-demand forecasting service. Using Google’s Vertex AI, Tchibo built a service called DEMON, which it can feed with more than three years worth of product, marketing, sales and logistics data. The service used a temporal fusion transformer model, like the transformer architectures responsible for large language models (LLMs), to predict online demand for its products up to 84 days in advance. The service helped the company manage its warehouse, reduce the time employees spend on logistics and see which products might be popular enough to bring back.
"We have been using our developments . . . to deliver reliable and rapidly scalable data and AI services at enterprise level for some time now,” says Marcel Knust, head of data science and AI, Tchibo.
Results
- Over six million predictions generated daily
- Reduced overstock and handling efforts saved significant time and money
- Reduced out-of-stock situations to support sales
Webinar
Nov
6
Fix the Content Bottleneck: Build a Better WebOps Strategy
Content stalled? Dev overloaded? You’re not the only one. Learn how streamlined WebOps bridges the publishing gap.
Register
Webinar
Nov
20
How to Build a Solid Knowledge Foundation for AI Success
See how leading brands keep their AI honest, compliant and actually helpful.
Register
Webinar
Dec
2
From Manual to Magical: How AI Transforms CX Teams
Learn how to replace manual support processes with automation that actually delivers.
Register
Webinar
On demand
Beyond Storage: Smarter Content, Bigger Impact with DAM + AI
Discover how the DAM + AI duo makes content smarter, stronger and more accessible.
Watch Now
Webinar
On demand
Agentic AI Playbook: Real-World Customer Service Use Cases You Can Deploy Now
Boost self-service by 30% and slash call volume by 63% with agentic AI.
Watch Now
Webinar
On demand
CMS Briefing: A Live Look at What’s Next in AI-Driven Platforms
Learn how leading organizations are using AI‑driven tools to publish faster, personalize smarter and stay secure.
Watch Now
5. Zavarovalnica Triglav
Zavarovalnica Triglav is an insurance group in Slovenia that found it needed a better way to handle customer inquiries. The company used disjointed communications channels, including emails and a call center, and it wanted to consolidate the routing of those queries, both to streamline processing and improve customer service, according to a case study.
Zavarovalnica Triglav first used Microsoft Dynamics 365 to centralize all of its interactions, including inquiries, sales and service support tickets. The platform allowed staff to see all of the communication and marketing involving a particular customer, so they could create personalized customer experiences. The company then went further, using Microsoft’s Azure OpenAI service to build a chatbot that could respond to customer questions. About 10% of contacts are common queries, and the chatbot provided answers to them, double checked by a customer service agent.
About 60% of queries cannot be handled with an automated response, so the company used Azure OpenAI to build a service that reviews their content and decides which department to route them to.
“We thought venturing into AI was a worthy step since it offers natural language. It’s what we would like to have with automation,” says Vladimir Nardin, director of the communication platform department, Zavarovalnica Triglav.
Results
- Manual intervention for certain types of requests cut in half
- Decreased manual workload due to automated transcription and summarization of communications
- Personalized responses to registered users from chatbot
See more: 5 AI Case Studies in IT
