The agentic AI space is seeing a surge in activity with disruptive advancements like OpenAI’s Operator and China’s Manus AI. These powerful agents can autonomously buy groceries online, book an Uber ride and trade in stocks through simple prompting. The emerging dominance of Big Tech in the space has begged the question whether agents are on the path to getting commoditised just like AI models — whether a swarm of agents will flood a marketplace and compete on price, and whether agentic startups will have any business moat to protect.

‍

AGENTS GO BEYOND LLMS

Large language models function as the “brain” for agents because of which agents can think, reason and function autonomously. While the brain remains central, each application/agent built on top of it executes a distinct task involving workflow integrations. Hemant Mohapatra, partner, Lightspeed, said while models are getting commoditised, agentic frameworks still have room for differentiation through UX, automation, determinism, and workflow integration. Homegrown startups are rapidly joining the AI race. In 2024, India’s total GenAI startup count grew to 240 from 66 in 2023, Nasscom data showed. Particularly GenAI assistants grew fourfold to 130 startups. Productivity- enhancing GenAI applications, especially coding companions and workflow augmentation tools, more than doubled to 45 from 20, backed by a twofold jump in funding.

“Many startups that function as model wrappers face existential risks unless they build proprietary data flywheels, domain- specific workflows, or deep integrations,” Mohapatra said. AI agents that go beyond chat and execute complex tasks autonomously will have the strongest staying power. As an investor, Lightspeed looks for world-class technical and GTM-oriented teams with data advantages, distribution, business translation, and real-world actuation capabilities, Mohapatra explained. “Indian startups should start rethinking where the value is going to accrue in AI,” said Arun Chandrasekaran, a distinguished vice president, analyst at research firm Gartner. “They should focus on developing AI solutions that address real- world challenges such as data analysis, workflow automation, and personalised recommendations to demonstrate tangible value to potential clients.”

‍

THE DIFFERENTIATION FACTOR

Vijay Navaluri, cofounder at agents startup Supervity.ai said agents must not be misunderstood as simply bots or tools which automate a task. “Having open-source developer toolkits in the market doesn’t mean that agents have become a plug-and-play solution,” he said. For sure, many startups will perish without a differentiating factor and a niche product market fit. “But organisations are undergoing an upgrade of business process automation techniques which requires domain expertise, data handling in multicloud environments, adherence to privacy and legal requirements, and most importantly talent,” said Navaluri. Supervity counts IBM, PWC and others among its go-to market and implementation partners.

Ashutosh Prakash Singh, cofounder at Bengaluru-based RevRag, said generic agents will no longer make the cut. “Vertical specialisation, domain expertise and early market dominance will be key for agentic startups who should act now and start building these market shafts,” he said. Backed by Cred founder Kunal Shah, RevRag specialises in AI sales agents which handle customer onboarding and KYC for fintech customers.

‍