Faced with a complex data environment and other challenges in finding data to quickly answer business questions, a multinational telecom company partnered with CGI to use GenAI to speed data-driven decisions dramatically.
The challenge
One of the largest telecom companies in the world needed faster access to the right data for decision-making. With over 183 million data columns spread across eight systems, managing data for reporting, building models, and answering ad-hoc queries required data analysts to spend countless hours on manual tasks, slowing operations and making crucial business decisions.
Proof of concept within 8 weeks
The company turned to CGI, a trusted partner for 30 years, to tackle these challenges and drive innovation. Together, they launched a GenAI R&D initiative to optimize operations and explore transformative use cases.
The client’s Chief Data Office (CDO) identified its top priorities as modernizing enterprise data platforms and delivering more intelligence and automation. The first phase of the project centered on unlocking the "art of the possible"—identifying the company’s vast data assets and determining how to use them more effectively.
Employing a lean approach, CGI provided training and expertise to help establish, expand and upskill the new R&D team. The joint team developed a working, lightweight proof-of-concept (POC) for a GenAI-powered text-to-SQL query generator, paving the way for more rapid and intelligent data management.
The POC featured a conversational bot, powered by Azure’s private LLM instance, to automate data discovery and SQL query-building. This innovation delivered explainable AI insights into how the model made its prediction and ready-to-use results for faster decisions.
Tasks that previously took from 30 minutes to 2 days were completed in just 45 seconds.
The joint team engaged with both leadership and colleagues to showcase the value of the POC model and generate enthusiasm for it. The team received funding to move the solution to production and scale, with additional CGI implementation support.
Phase 2: Production and scale within 4 months
Next, the telecom company and CGI worked in agile teams to deliver the solution, tapping CGI’s global delivery center to access the on-demand expertise needed to scale.
A modular microservices design ensured flexibility and reusability across teams. If one team needed to verify access to a specific dataset, a modular authentication component could be seamlessly integrated for reuse, accelerating deployment.
The teams also used CGI’s pod delivery model, which co-locates research, development, testing and other experts with diverse and deep skill sets. This pod structure became essential during solution engineering to foster collaboration and increase efficiency. It also allows the company to tap into an experienced delivery team of AI experts for future initiatives without creating separate teams.
Further efficiencies were gained by integrating Agentic AI. AutoGen directed incoming questions to the right data sources, and from there, a team of AI agents worked together to find the best answers. Some specialized in learning from examples, while others focused on identifying and correcting mistakes. This collaborative approach allowed the team to achieve the highest accuracy, with built-in error correction to improve results automatically.
Self-service intelligence tool delivered in just 4 months
Using ChatGPT, the full solution provided a powerful tool for self-service business intelligence. It has revolutionized how employees across the company navigate the data and AI life cycle, from finding and accessing data required for machine learning and reporting to creating visuals on the fly to drive business decisions.
Now, analysts and users can quickly respond to ad-hoc business queries such as: How many former fiber customers do we have? How many active corporate executive customers do we have? How many wireless postpaid subscribers have a tablet device?
Business intelligence is integrated across the enterprise, providing immediate insights to make AI-enabled decisions at scale. This capability has bridged the gap between business and IT teams, enabling a collaborative focus on strategic priorities. What’s more, CGI’s GenAI partnerships with alliances such as Microsoft have promoted resource sharing and optimized development costs.
Tangible outcomes at a glance:
- Increased operational efficiency: Faster, better data discovery and SQL query-building through automation.
- Enhanced decision-making: Improved data accuracy, speed and AI-enabled decisions across the enterprise.
- Empowered employees: Increased data accessibility and transparency, putting powerful data analytics capabilities into the hands of every business leader and user.
- Innovative culture: Boosting the ability to innovate data platforms, the supply chain, and the science ecosystem at scale.
“We have a vision to make data more discoverable for our users and need data engineers.”
- CDO Architect
Laying the foundation for an enterprise GenAI platform
This solution delivers value by creating modular agent frameworks that can be reused across various GenAI use cases, accelerating production with common, reusable components. It also lays the foundation for a scalable GenAI enterprise platform.
Partnering for ongoing success
The telecom client will continue executing the product roadmap to further unleash the potential of GenAI solutions, backed by its modernized and automated platform and with CGI as its trusted partner. The journey ahead looks promising, as it will further optimize operations and customer service to differentiate the company from competitors.
The solution’s robust framework has been recognized globally at the top of the GenAI leaderboard, which measures how well large language models can parse a question in natural language and translate it to SQL.
Apprenez-en plus à propos des capacités, de l’expérience et des résultats en matière d’IA de CGI.
