AI is disrupting software building and creating competitive pressures, while creating new opportunities for SaaS (Software as a Service) companies, says Murali Swaminathan, chief technology officer of Freshworks. The company is adapting to this shift, changing how it designs, engineers and develops products. In an interview, he discusses how agentic AI is deployed in large enterprises and how youngsters entering the job market should approach skilling. Excerpts:How is AI changing SaaS?AI is no longer altering how people use software, it is changing how software is built and how work gets done. AI makes building easier and faster. Now that everyone can be a builder, there is increasing competitive pressure and we are getting disrupted. But that is a positive, as we are learning and transforming. Competitors make us stronger, and we are now able to compete with big players serving large enterprise customers. Freshworks is adapting its internal product development processes and changing how our teams think and build.What changed the way you build software as you target upmarket customers?Core engineering fundamentals do not change, but enterprise expectations and business approaches are different. Massive enterprise deals demand higher system resiliency and urgent support. The go-to-market motion (how we sell and deliver the product) shifts from anonymous, self-serve usage to a lengthy, high-touch relationship. This requires sharing our future roadmaps in advance, offering complete transparency, engaging directly with executives, and providing hands-on trials.How can companies like Freshworks differentiate?While AI makes building initial software prototypes fast and effortless, the challenges remain- scaling applications, ensuring system resilience and maintaining consistent operations across thousands of enterprise users. So, the conversation has shifted from merely experimenting with AI to operationalising it safely and reliably at a massive scale. Technology is rapidly advancing beyond basic chat assistants toward networks of agents and fully autonomous systems. It is a decade-long journey with vast opportunities as AI-driven software transforms how complex work gets done. We are moving from simple AI copilots that just assist users, to AI agents that can execute multi-step tasks, invoke tools and take real actions.As SaaS companies move towards automating workflows, what security measures are required?I see enterprise software evolving into a trusted system of record, context and control. As AI agents move from just answering questions to taking actions, safety guardrails cannot live only inside the AI model itself. An enterprise agent needs an identity, defined permissions, and a clear scope of authority much like a human worker does. An agent summarizing a document is different from one changing critical production infrastructure or modifying system access. As the potential business impact increases, so should verification and oversight. This is where mature SaaS platforms remain vital. They provide systems of record, APIs, workflows, role-based access, and audit trails. Even if users increasingly interact through AI agents instead of traditional screens, these agents still need that trusted foundation. The real advantage is combining trusted context, identity, permissions, and controls.How are you adopting agentic AI?A lot of emphasis on adopting AI agents in enterprises focuses on governance and security, which includes giving these agents role-based access similar to human workers. We are introducing an entity called a non-human agent or “digital worker”, AI agents acting on an employee’s behalf. These digital workers are treated as personal identities with scoped, time-bound permissions. For example, an employee could temporarily delegate authority to their digital worker while on vacation. To maintain enterprise governance and accountability, system logs must explicitly distinguish between actions performed directly by a human and those executed by their digital worker.What is your advice for freshers?AI is only as good as the person using it. As basic coding tasks are increasingly automated, entry-level expectations are higher now. Youngsters must use AI to delegate routine work so they can focus on solving more complex problems. Crucially, they must understand core computer science and programming fundamentals to judge whether AI-generated code is correct and challenge it when necessary.How should they approach education?While catching the AI wave early is valuable, completing a college degree remains essential. It provides structured learning, human experience, networking, and a broad foundation that self-taught building alone cannot replace. Candidates are expected to present a portfolio of things they built, a simple app for the family, a cloud deployment, or a small tool. Employers expect candidates to know basic platforms.
