AI is Killing the 5-Year IT Project

The traditional IT services playbook isn’t dying; it’s simply being rewritten in code no human can read alone. For firms like Innover Digital, the disruption GenAI has brought is not about replacing engineers or eliminating services. Instead, it’s about replacing the way those services are delivered.

Once dominated by hundred-member integration projects and five-year SAP rollouts, the system integrator (SI) model is being unbundled. What’s emerging in its place is smaller, smarter, AI-first pods, where human developers work alongside agents that read, write, test, and deploy.

Rakesh Prasad, senior VP of strategy and solutions at Innover Digital, told AIM, “We are infusing AI into everything that we do. It’s not that I’m changing and becoming an AI company only…It is how I deliver those options.”

The Old Model Didn’t Die—It’s Just Shrinking

Innover Digital, a digital engineering firm, is now orchestrating AI-native delivery models to provide benefits over traditional software delivery methods. 

Before GenAI, software engineering delivery revolved around well-staffed pods. One such example Prasad shared included AI-enhanced pods that replace a seven-member delivery team with a combination of bots and humans.

“Now, we are saying that the pod model has to come in, with AI agents as part of the pod, which means I will give you only one product owner, maybe one good developer, and two AI pod code writers, one QA agent, one deployment agent, and one DevOps engineer,” Prasad said.

This restructuring doesn’t remove people; it repositions them to work more efficiently with three humans and four bots. 

The output remains the same, but the ratio has changed. Prasad believes that it is not that software engineering is not being done; it is just how companies like theirs deliver software engineering and charge commercially.

Goodbye, 5-Year Implementation Plans

The kind of projects that once defined the IT industry are, in Prasad’s words, fading. “There was a time, when SAP implementation used to run for four, five years—one implementation of about 100-120 million dollars. Can you imagine an IT project today running for five years? And nobody even batted an eyelid.”

That model was fuelled by enterprise-scale software complexity, but is now being undercut by cloud, SaaS, and GenAI. 

“Your traditional SI work—putting 100 members to build integrations and make one data move from another and build the screens—that is gone,” he stated. 

This isn’t just faster, it’s also necessary. “There will never be one big application that will do everything for a company. The need is still there, but the need is changing to be small, more intelligent, and more high-end work,” Prasad said.

Legacy Isn’t Dead, But It’s Finally Getting Mapped Using AI

Even as new platforms emerge, Prasad is clear that legacy systems aren’t going away. 

“I think that change is very imminent, how you work with legacy, will be very critical, because not all [things can be modernised] quickly, it’s very hard to modernise. While it sounds very fascinating, the mainframe is still, in airlines, banking, and runs at the core.”

He believes regardless of how often the demise of mainframes has been declared, they persist. Legacy systems will always exist. The key is to find its purpose before modernising it.

Instead of forcing a rip-and-replace approach, he believes the solution is to build intelligent integration layers.

He provided an example of Innover Digital helping a financial services firm stuck with a 7,50,000-line legacy ASP.NET codebase using a SQL server. They have a couple of bots built on a software engineering lifecycle and agents aimed at legacy modernisation. Moreover, the company is using these solutions to decipher code and reverse engineer to document things.

“Every legacy system will throw a different challenge…so your ability to figure out how that last mile interaction will happen is important,” Prasad noted.

The Future is Smaller, Smarter and Far More Targeted

Perhaps the biggest shift is philosophical. Innover believes in not building bloated, one-size-fits-all agents. Instead, every solution should be surgically applied—tailored to a specific process, client and bottleneck.

“I would say we are lucky. I don’t want to claim we were very intelligent. We are just lucky because we are smaller, so we can make that shift,” he said.

“Most of the tech stack in the legacy, I think, is on that Java and TypeScript side…We’ve not built any custom .NET-based applications in the last two to three years,” Prasad noted. 

GenAI isn’t replacing IT services. However, it’s steadily replacing the assumptions on which they were built. There are no five-year plans anymore, just quick integrations, AI agent-powered pods, and the slow death of the old services blueprint.

The post AI is Killing the 5-Year IT Project appeared first on Analytics India Magazine.

Generated by RSStT. The copyright belongs to the original author.

Source