Stack Overflow on AI: A Computer Weekly Downtime Upload podcast

Heading a company that is focused on software developers, Stack Overflow’s CEO, Prashanth Chandrasekar, is probably in a fairly good position to spot trends in software engineering. Today the company is seeing a rise in the use of AI for coding. Chandrasekar’s view is that the code being produced by these AI agents is not comparable to what can be achieved by an experienced software developer. It may not even be elegant, but he feels that an AI agentic can do the job well enough that a human coder is not needed.

In answer to a question exploring the industry hype associated with agentic AI in the enterprise, Chandrasekar says: “If you even just think about coding, what we noticed is that digital workers or agents autonomously go and perform tasks to achieve the outcome, irrespective of how they do it, which I think is somewhat secondary.” 

“AI agents may produce inelegant code; they may produce things that are not exactly perfect compared to an expert developer, but it gets the job done in a fraction of the time.”

Chandrasekar believes AI fundamentally changes the way to accomplish tasks in an organisation. He says: “I think roles completely change. It’s more fundamental than, let’s say, a platform shift, in my opinion. I think it’s definitely a platform shift, but agentic AI does a lot  more than just that.” He’s a big fan of Star Trek and sees agentic AI as a fundamental energy that has been infused across the business world. “It’s a complete sea change.”

As an example, he expects product designers, product managers and engineering jobs will change. “What is the role of a designer, a product manager, or an engineer? I think that will change.” For Chandrasekar, this change is analogous to the transformation cycles that have occurred previously, but AI is also different. “It is similar to former cycles, but I suspect it’ll go faster, relative to prior cycles, due to the diffusion of the technology.” In effect, because AI is being infused into every piece of enterprise software, it will touch more people more of the time and, as such, everyone’s way of working is set to adapt to the new capabilities the technology offers.

Whatever the size of the organisations, Chandrasekar believes workflows that are agentic AI -enabled  can be engineered in a way that is logical, where they understand context and are connected to other workflows to achieve the desired outcome. “I think this is a very powerful mechanism,” he adds.

However, the challenge the majority of IT leaders face is technical debt, which continues to grow and needs to be maintained. An example is robotic process automation, which gave enterprises a way to work-around pain-points in business processes caused when workflows require manual intervention.

Given many organisations have yet to automate systems fully, they appear to be playing catch-up with a tech sector that has already moved onto the next big thing, which today means agentic AI. Chandrasekar believes agentic AI represents a phenomenal opportunity in enterprise IT: “The agentic revolution today, where people are initiating or initialising these AI agents to operate on behalf to perform various tasks, is quite profound. These agents will perform actions on your behalf in various capacities,” he adds.