In recent years, Generative AI (GenAI) has grown in leaps and strides, finding applications in different fields, especially software development. Integrating AI and GenAI into the software development lifecycle comes with numerous benefits, enhancing every phase from concept to deployment. From automating repetitive tasks to streamlining workflows and reducing errors, these technologies are redefining the software development process.
However, the emergence of new tools every day and the seemingly endless number of use cases can be overwhelming for developers to navigate the GenAI landscape. At DevSparks Bengaluru 2025, Rahul Bhattacharya, AI Leader, EY GDS Consulting, explored how developers and engineers can apply AI-powered tools to transform business applications in real-world tech environments.
Here are the highlights from the session.
Building a problem-first mindset
The vast number of AI-powered predictive and generative tools available to developers today is both a boon – and a bane. The sheer variety is sometimes overwhelming, and finding the right tool is often a challenge. However, Bhattacharya recommends finding the problem before choosing the tool. “Problems,” he said, “exist independent of solutions. Look for the right tool to apply to solve the problem. So, always start from the end goal and work backwards to find the tool.”
GenAI tools: reward or risk?
The discussion also tackled the ethical uses and risks of leveraging these tools. While Bhattacharya notes that ethical considerations may not necessarily apply to coding, he highlighted the risk of using copyrighted code in training large language models (LLMs).
Organizations need to contend with the risks of opting for AI/ML tools. “Think about your appetite for risk within the enterprise. What is the possibility that, even if there is a 1% chance, what would be the cost of that outcome? Because risk is not just a probability. It is also the impact of something going wrong. If you remember probability and statistics, you will recognize the concept of expected value (the probability weighted value of any random event). So, it’s important to look at that and decide: Is it worth the risk?” he said. Risk requires a case-by-case evaluation.
He cited the commercial aviation industry as a powerful analogy. Despite 80% of in-flight operations being automated through advanced control systems, sensors, and actuators, the stakes are so high that every flight still requires two pilots to oversee the journey. “Every year there are a dozen papers about the feasibility of autonomous aviation, but we’ve not even gone from two humans to one human in the cockpit,” Bhattacharya noted. At this point, he believes risk evaluation is key when it comes to leveraging AI/ML tools. Organizations must carefully assess the level of oversight and human control required in an AI-driven system and proceed accordingly.
Driving innovation for customer service
Bhattacharya also discussed how EY GDS – an innovation hub and technology powerhouse within the global EY organization – provides business solutions to member firms and is applying AI/ML tools to address client challenges and drive innovation. He explained how AI-powered solutions at the global EY organization are being applied across the enterprise – from the front office to backend operations and the middle office. The front office, which typically handles customer care, customer service, operations, and sales, requires a different set of solutions.
Take the example of contact centers. “When customers call in, they often hear that the conversation will be recorded for quality assurance purposes. In reality, only 5% of those calls are actually reviewed as it’s just too resource intensive for human teams,” Bhattacharya said. The global EY organization addressed this by integrating GenAI tools that now evaluate 100% of customer calls.
From these results, organizations can create personalized training for individual agents, sharing critical feedback to improve the customer experience. Additionally, contact centers typically spend 15 minutes after each call logging notes on customer satisfaction, follow-ups, etc. With GenAI tools, these tasks can be streamlined with automated summaries generated, reviewed, and submitted with minimal manual effort.
Streamlining contracts for middle offices
Middle offices handle supply chain, inventory, processing systems, contracts and other departments that typically involve heavy paperwork, said Bhattacharya. He pointed to large telecommunication companies that function through contracts and contract management.
To address inefficiencies and reduce risk, the global EY organization introduced AI-powered tools that can extract key entities, check for compliance to standards, and spot differences across multiple contracts, thereby improving the value of the agreements, reducing risks, and help enable compliance.
Erasing the drudgery of data lineage
The back office, which houses the developers, IT system management, and more, is ripe for disruption with GenAI and AI tools. One of the common projects the back office handles is cloud migration, particularly the shift from on-premise infrastructure to the cloud. The organization recognized that in this process, establishing data lineage is extremely important. Typically, this task has been manual and prone to errors: employees read through the scripts, trace data sources and record it in a spreadsheet, and then repeat the process ad nauseum. To streamline this process, the organization introduced GenAI systems so that the first round of this process is automated.
“A lot of the work we do atEY GDS is enterprise transformation projects. Data migration is a large part of the process. Finding the data lineage, and doing source to target mapping? Not interesting or technically challenging. It’s drudgery. All of this can be automated and make us more productive by taking away the low value-add work,” he shared.
EY GDS has made it such that developers do not even need to write a prompt to automate these processes – the prompt is engineered and placed within the IDE.
Bhattacharya discussed a range of use cases for predictive and GenAI tools, including one in advertising. By using multimodal models and computer vision, the global EY organization was able to reduce a process from 45 minutes to a mere 45 seconds, with all brand guidelines in the ad being followed. He concluded the session with a call to developers to embrace AI, invest in learning, gain experience, earn relevant certifications and take the leap into India’s AI-driven workforce.
Original Article
(Disclaimer – This post is auto-fetched from publicly available RSS feeds. Original source: Yourstory. All rights belong to the respective publisher.)
