What software development use cases is GenAI currently best at, and where does it still fall short?

A significant challenge at the executive level is identifying use cases within AI that can enhance customer value and increase revenue. While GenAI optimizes many processes, it hasn't yet delivered business-specific use cases that add substantial value. This might be because most training has been done with external data. As we move into 2025, training with internal business data could uncover new use cases that boost revenue and business value.

We're cautious about large-scale efforts with GenAI. It's effective at generating and summarizing small code snippets, but not for developing entire applications. GenAI struggles with larger contexts and tends to hallucinate or invent details in extensive code generation. However, it's quite adept at creating small methods, functions, tests, and reviewing documentation.

GenAI can operate across the entire software development lifecycle. Over the past year, we've conducted pilots, rollouts, and data gathering to see where it fits best. It excels in code generation and understanding, especially with legacy code. There's significant value in these areas, and we're exploring how GenAI can enhance customer experiences and journeys. By integrating agents into existing workflows, we're rethinking how problems are approached. While GenAI can do a lot, its effectiveness depends on governance, trust in data, and other enablers.