Agentic AI is reshaping tech and trust in India
Amidst the excitement of everything AI, agentic AI systems offer perhaps the most tangible peek into the potential of this technology.
At CES earlier this year, Nvidia CEO Jensen Huang raised a lot of eyebrows when he declared AI agents a multi-trillion-dollar opportunity. While current AI tools excel at responding to our prompts, agentic AI systems promise something bigger: the ability to independently understand, plan, and tackle complex tasks.
So what exactly sets Agentic AI apart?
Agentic AI is Independent, Adaptive, and Collaborative
Whenever someone says “agentic AI”, the first thing we think of is complex AI models running on state-of-the-art tech, doing god’s work — but we often don’t realize that Siri and Alexa were the original AI agents.
Let’s break it down:
Think of traditional AI as a highly skilled assistant – it's great at specific tasks like analyzing data or finding patterns, just like how Netflix’s recommendation engine is great at finding patterns and suggesting what you should watch next. Generative AI took this further. It became the creative colleague who could write, draw, or compose based on what it learned. For example, you could ask ChatGPT to write a poem for you in Jay-Z’s tone or ask Dall-E to generate an image reflecting India’s diverse landscape, but it still needed our prompts to get started or improve the content.
Agentic AI is more like having a proactive assistant who doesn’t wait for instructions but identifies what needs to be done and handles it themselves. It also learns from its results to do better next time, it adapts its course without needing a human to step in.
These AI agents can also work together in multi-agent workflows – Anthropic did a great overview recently. Imagine a team of specialists collaborating on a project: one researcher gathering data, another crunching the numbers, and a third writing up the findings. That's exactly how AI agents work together. They share information through a common communication channel, and each handles their specialized part of the task.
These higher tiers include specialized agents that interact with users, while governance agents act as the system's conscience, ensuring everything stays within ethical and legal boundaries - almost like a smart corporate structure. Here’s an overview of a hypothetical investment firm run by AI agents.
Acting as a catalyst for rapid innovation in Indian startups
In just a few years, Agentic AI has evolved so rapidly that many of today’s startups may find their offerings obsolete within six months.
It all started with horizontal agents that could do everything, yet nothing. Siri could send you a reminder for your meeting but not make contextual notes of the meeting. Alexa could turn on the lights if you asked, but it couldn’t adjust the lighting per the temperature or ambiance. These agents also mostly understood English, and while they claimed to understand a few vernacular languages, “Sorry, I couldn’t understand you” is all you’d get.
The next generation of agents came in the form of vertical agents who could only do a few niche tasks but did them faster and better than anyone else. We saw this with Github’s copilot first. These agents work great with privacy and trust since they usually operate on an on-premise basis.
Now, at a nascent stage, we’re seeing many startups building workflows wherein different agents solve for different parts of the chain. These are a combination of ‘if-else’ loops wherein different agents get triggered based on conditional statements. We’re also witnessing startups simplifying the process of building an agent or a workflow, almost making the whole thing into a no-code tool.
In the Indian context, one of the biggest breakthroughs lies in vernacular language support.
Indian startups are already capitalizing on this opportunity. We’ve seen AI agents designed for niche but critical applications like loan collections, particularly targeted at rural areas. For example, these agents can remind borrowers of repayment schedules, explain loan terms in their native language, and even handle escalations with tact, all without the inefficiencies of manual intervention.
This trust-building aspect is especially crucial in a country like India, where technology is often met with skepticism. When AI systems can communicate in familiar ways, operate transparently, and deliver results reliably, they begin to earn the trust of users — be it a farmer in Madhya Pradesh, a shopkeeper in Gujarat, or a student in Meghalaya.
What next?
The impact of agentic AI, by all means, is inevitable.
The dawn of the “agent economy” will see networks of autonomous AI assistants with the potential to take on most grunt office work. That will free us to focus on what truly sets us apart: building relationships, developing trust, and applying creativity.
It’s not just Nvidia; the tech industry's biggest players are all in. Microsoft's Satya Nadella recently launched a dedicated AI division focused on agentic capabilities, while Google, OpenAI, Salesforce, and AWS are also investing heavily in this space.
These systems are quickly evolving to become problem-solvers, not just mere responders.