BREAKING
PM Modi to address IIT Delhi convocation, inaugurate AI-powered supercomputing facilityNon-stop Rain Brings Delhi-NCR To A Standstill; IMD Forecasts More Showers And ThunderstormsMandaviya calls for transparent athlete selection, stronger sports governanceHeavy downpour leaves Delhi submerged as traffic crawls, infrastructure faltersKerala intensifies search for 3 fishermen missing after separate boat mishapsDurand Cup: Rodriguinho scores five as SC Delhi thrash Defenders FC 7-0AI to Play Bigger Role in Making India’s Bus Transport Smarter: UITP ConferenceIndia aware of delays in Canada visa processing for retired officials: CentreGolconda Gladiators, Secunderabad Stallions clinch Hyderabad District Basketball Championship titlesUS Currently Provides 67% Of India’s Imports Of LPG, Says Hardeep Singh Puri
Technology

Zoho CEO Says SaaS Is Simple, But AI Agents Aren’t

hellohimachalse@gmail.com 07 Aug 2026 1 hour ago

The AI industry is heavily invested in AI agents. They are touted as the ultimate digital workers, capable of managing tasks, making decisions, and completing workflows with far less human intervention: in short, more work, higher productivity, and less need for labor. Indeed, the rise of AI agents like Claude in early 2026 shook the SaaS sector, causing a drop in company stocks and a wave of layoffs. However, while AI agents are hailed as the future of work—smarter and more capable—Zoho CEO Mani Vembu doesn’t think it’s that simple. For Vembu, SaaS remains easier to use than AI agents.

In response to a post on X by Josh Miller, CEO of The Browser Company—who argued that its surprising AI agents haven’t yet become a mainstream phenomenon—Vembu offered a different explanation for their slow adoption.

Miller pointed out that AI models, agent frameworks, and tool invocation capabilities have advanced rapidly; however, most people outside the tech industry don’t seem particularly interested in using AI agents. “These cutting-edge AI models are amazing (as are the control systems, tool calls, and so on). Plus, every major tech company has an AI agent platform, not to mention the Y Combinator startups developing vertical agents,” he wrote.

“Yet your friends and family outside the tech world—who spend their days staring at their iPhones and get paid to work in browser tabs—don’t really care about or use any AI agents yet.” “Most people still use ChatGPT and Claude as an improved version of Google and Grammarly,” Miller added.

Miller questioned why AI agents haven’t achieved the massive success of products and services like Instagram, TikTok, and Uber.

In response to his post, Vembu agrees that AI agents haven’t yet reached the mainstream, but argues that the problem isn’t so much whether people want them, but rather the amount of work required to make them truly useful. “The problem isn’t the models. It’s the time it takes to generate value,” Vembu wrote.

“SaaS delivered value almost immediately thanks to self-service. AI agents require an ‘agent supervision’ phase before they become useful. You have to teach them about your business, your data, and your workflows,” he added.

According to Vembu, this is one of the biggest differences between traditional SaaS and AI agents. The Zoho CEO points out that while SaaS products can typically be used almost immediately after signing up, AI agents require setup and training on how the business operates before they can provide significant value.

“Even for the same use case, the same agent rarely works right out of the box without prior adjustments for two different companies,” Vembu wrote.

“Until onboarding an agent is as simple as signing up for a SaaS service, mass adoption will take longer.”

Interestingly, this isn’t the first time Vembu has highlighted the challenges of deploying AI agents. In a previous in-depth post, he argued that enterprise software is moving from simply helping employees perform their tasks to replacing parts of the work itself with AI agents. “AI agents are different. They are no longer just tools.” “They are expected to perform parts of the work that humans previously did,” Vembu wrote.

He pointed out that, unlike traditional software, AI agents need to learn about a company’s products, processes, policies, and workflows to become effective digital employees. “Every company trains its employees differently because every business is different. AI agents require the same training,” he stated.

Consequently, Vembu argued that deploying AI agents will require much more ongoing training than traditional SaaS. “Implementing AI will require much more training than SaaS. The software may be the same, but the knowledge is not,” he wrote.

Related News