Inside large IT firms, software projects used to begin with a deck and were heavy on documentation. Then came the shift towards Agile methodologies, still largely followed to date, that focus on iterative development.
However, technology services company Hexaware Technologies believes that the cycle is ripe for reform. Several IT firms have already announced AI and coding partnerships, using vibe-coding tools including Lovable, Cursor, and GitHub Copilot internally. Hexaware is extending its partnership with Replit and wiring it directly into its internal engineering platform, RapidX.
As a part of this announcement, Hexaware has also established a dedicated Centre of Excellence for AI and software development life cycle at its Chennai campus, to boost AI development and help increase adoption.
The company calls it ‘business vibing’. Business teams express what they want. IT builds something live in minutes. Then engineering takes over and hardens it for production.
By bringing RapidX and Replit together in a single pane, the company says enterprises can now “validate ideas earlier, shorten decision cycles, and scale successful prototypes into enterprise-grade systems with stronger engineering discipline and governance while empowering non-technical knowledge workers to participate directly in the software creation processes.”
For a services company that runs on large, multi-year deals, this is a different posture. Sanjay Salunkhe, President and Global Head of Digital and Software Services at Hexaware, frames it as a compression problem.
“In the AI-driven life cycle, one of the biggest crunch which is happening is in the intent and the requirement engineering phase. ‘How do I quickly get to a point where everyone agrees’—this is what I want to build,” he tells AIM.
That first step, he argues, is where most projects stall.
Vibe Engineering
Hexaware positions RapidX as an answer to that friction. It is a GenAI-native software engineering platform that helps teams understand legacy code, blueprint new systems, and push forward engineering.
Now, Replit is the new front-end.
A business user sketches an idea. Replit spins up a working app. RapidX then translates that into architecture, standards, and production code.
For Eravi Gopan, President and Global Head of Technology, Products, and Platforms at Hexaware, this is less about tools and more about timing.
“Replit has been a rapid growth star of the AI era,” he says. “We’ve been an early partner. We’ve been customer zero, using that for internal transformation as well as part of our global market process in a very tangible way. We created solution sets using Replit. We use that for our prototyping, our demos, our solutioning for our customers.”
Hexaware has been using Replit across dozens of engagements. It earlier partnered with Replit on its latest release, Agent 3, which launched globally in September last year. Hexaware had early access and shared feedback during the product cycle.
The integration is designed to be tight.
“When I’m now doing a modernisation programme, and I’m changing the business workflow, Replit creates a quick mock. The business goes and endorses it, this is what I want. IT understands what it wants, and RapidX actually implements it through the AI-driven SDKs,” Salunkhe explains.
In practice, this plays out across two types of work that dominate enterprise IT: greenfield, where a company launches something new, and brownfield modernisation, where decades-old systems need to be reformed without breaking operations.
Hexaware also hosted HackaSprint with Replit, an invite-only evening meetup allowing attendees to build software with vibe-coding and enabling builders who have never written a line of code to turn ideas into working apps.
The Internal Solution
The harder problem sits between business and IT. Business speaks in outcomes, IT works on tickets, and a lot is lost in translation.
Salunkhe believes vibe coding bridges that distance. Hexaware is now forming what it calls RapidSquads—small teams that sit with business users, build live prototypes in Replit, and either push them into RapidX or hand them over to core engineering.
“It quickly empowers functional consultants [and] solution architects with the context they already have. Earlier, they would depend on five or six developers. Now they can create [themselves],” Gopan notes.
The partnership also links to another bet Hexaware is making, something it calls ‘Zero Licensing’.
The thesis is that enterprises have overbought software for years. Multiple SaaS tools overlap, with heavy customisations and large subscription bills that never quite go away. Instead of stacking more licenses, Hexaware wants clients to build more themselves.
“Build has become cheaper and easier,” Salunkhe explains. “Things which you’re buying for years to license setups, you’re customising the hell out of it. Eventually, you are spending an equivalent amount of what you’re doing with the build, and then there is a lock-in with that particular vendor. GenAI has made this whole process much faster. You can now really code what you want in a much faster way.”
Replit handles quick creation. RapidX handles enterprise guardrails. Together, they can replace layers for smaller packaged tools that once sat on top of core systems.
Critics might argue that quick builds create more technical debt. However, Hexaware doesn’t believe so.
“In fact, I think it’s an antidote for technical debt,” Gopan remarks. “Companies are already carrying one or two years of tech debt across different parts of their stack. This actually should do away with it.”
Behind all of this sits a more pragmatic goal: revenue. Indian IT services firms have started carving out AI-led numbers. Gopan does not pretend everything will suddenly become agentic.
“It’s never going to be one thing,” he adds. “Some systems will continue in a very traditional mode. But rapid application development using AI engineering opens up a bunch of opportunities.”
Shorter cycles also change the deal math. A project that once took 3 years can be squeezed into 10 months. The spend does not disappear. It shifts and often concentrates.
Salunkhe observes that, internally, AI is already everywhere. “Almost all our employees are touched by AI. It has become very native to us now. Almost all programmes we evaluate use to be done by AI,” he notes.
Your reaction
Discussion
What's your take on this story?
No comments yet — be the first to share your view.