Rajesh Jain on Agentic Marketing, Human Judgment, and the Speed of Belief

The internet pioneer and Founder and Managing Director of Netcore.ai discusses what agentic marketing means for brands, where human judgment remains essential, and parallels between the dot-com era and today’s AI boom.

Rajesh Jain

There is no shortage of enthusiasm around AI, but Rajesh Jain knows something about technology booms. He founded IndiaWorld Communications, one of India’s earliest internet portals, in 1995 and sold it four years later for $115 million, a major internet deal in Asia at the time.

Today, Jain is Founder and Managing Director of agentic marketing platform Netcore.ai. Though a long way from the portal days, he sees parallels between the internet in the late ’90s and AI today. The one that strikes him most, he says, is “the speed of belief, not the speed of the technology.”

We recently caught up with Jain to discuss what agentic marketing means for brands and why Netcore became Netcore.ai earlier this year. He also talks about AI autonomy, customer trust, and his advice for marketing teams looking to incorporate AI into their strategy for the first time.

advertisement

Earlier this year, Netcore rebranded as Netcore.ai. What was behind the decision, and what does the change represent for the business?

Everyone in martech added “.ai” to their name this year. Almost nobody rebuilt the platform underneath it first. We could have changed our name two years ago. It would have been a lie.

The question we kept asking ourselves internally was simple: are we actually an AI company, or a company that added AI to what we already had? We didn’t want to be the second one, so before we changed the name, we changed the architecture. Customer data, engagement, and channels, which had run as fragments for most of the industry’s history, came onto one platform, with a single context layer every agent draws from and one governance layer applying the same consent and brand rules across all of it.

Netcore.ai isn’t a rebrand in the usual sense. It’s what we became after we did the work, and the name just followed.

advertisement

For those unfamiliar, what does “agentic marketing” mean for marketers and brands?

Most of what gets called “AI in marketing” today is still automation with a recommendation bolted on: a human still decides the segment, writes the content, and hits send; the AI just makes one step in that chain a bit smarter. Agentic marketing is different in kind, not degree. An agent plans, executes, analyses, and optimises a customer interaction without a person sequencing every step, inside guardrails a human has set in advance.

Most of what gets called ‘AI in marketing’ today is still automation with a recommendation bolted on: a human still decides the segment, writes the content, and hits send; the AI just makes one step in that chain a bit smarter.

The system decides. The human aims. That’s the whole distinction, and it’s a bigger one than it sounds, because it changes what’s actually possible at scale, personalising for every individual customer rather than the handful of segments a human team can realistically manage.

Netcore.ai ties 30% of its platform fee to customer KPIs. Can you talk more about that, including what the thinking was behind that model?

In selected engagements, where we can establish a credible baseline, 30% of our platform fee is tied to the client’s own business KPIs: retention, lifetime value, the outcomes they actually hired us to move. The thinking behind it is simple: for fifteen years, this industry has priced itself on inputs, messages sent, monthly active users, licences, because inputs were the only thing anyone could reliably measure at scale. That model lets a vendor get paid regardless of whether anything actually improved for the client.

Once a platform can act on a complete, current picture of a customer rather than a fragmented one, there’s no excuse for that anymore. We wanted our own incentives to sit on the same side of the table as our clients’, not across from them.

You built IndiaWorld during the dot-com era. What parallels do you see between that period and the current AI boom, and what feels different this time?

The parallel that strikes me most is the speed of belief, not the speed of the technology. In 1999, capital and attention moved toward the internet faster than most people’s mental models of what it was actually good for. We’re watching the same thing happen with AI: the deployment is outrunning the understanding, in both directions, companies overestimating what it can do unsupervised, and underestimating how much unglamorous groundwork it needs before it’s trustworthy.

What’s genuinely different this time is the self-awareness. The dot-com era had almost no serious conversation about its own risks while it was happening; that came later, in the crash. This time, the people building the technology are having that conversation in public, in real time, before the crash rather than after it. Whether the industry actually acts on it as fast as it talks about it is the open question.

What’s your take on recent calls from some of the industry’s biggest names for greater restraint in AI development and deployment?

I’d separate two different conversations that are getting collapsed into one. What Amodei [Dario Amodei, co-founder of Anthropic] and others are raising is about frontier model capability, systems that may eventually operate with a level of autonomy nobody fully understands yet, and that’s a genuinely different risk category from anything I work on day-to-day. I don’t think it’s my place to referee that debate; I’m not building frontier models.

What’s genuinely different this time is the self-awareness. The dot-com era had almost no serious conversation about its own risks while it was happening, that came later, in the crash.

What I’d say from where I actually sit is that the instinct underneath the call, that capability should never outrun the ability to govern it, is exactly right at every scale, including the much smaller one I operate at. Every agent we’ve built works inside guardrails a human sets, and there are decisions it’s never allowed to make on its own, regardless of what the data suggests.

That’s not caution for its own sake. It’s the reason the system can be trusted with real decisions at real volume. If there’s a lesson from the frontier-model conversation that applies directly to marketing, it’s that: build the governance at the same pace as the capability, not after it.

 

As AI takes on more marketing decisions, where does human judgment remain essential?

Two places, specifically. First, setting the goal and the guardrails in the first place, what the system is optimising for, and what it’s never allowed to do regardless of what the numbers say. An agent can be excellent at finding the best path to a goal and still be aimed at the wrong goal entirely; that’s a human decision, not a modelling problem.

Second, anything financially unusual or emotionally sensitive for a specific customer at that moment, a discount well outside the normal range, an interaction close to a complaint or a churn flag. Volume alone isn’t the trigger for human review. Those two conditions are.

Recent research suggests marketers are adopting AI faster than consumers are becoming comfortable with its use by brands. How should marketers navigate that gap without undermining consumer trust?

That gap is real, and I think the honest response is to slow down exactly where a customer would feel unsettled by discovering AI was involved, not where it’s technically capable of moving fast.

Adoption speed should be set by customer comfort, not by what the technology allows.

Anything that touches money, health, or a moment where a customer is already upset deserves more transparency and a lower threshold for human review, regardless of what the system could technically handle on its own.

Adoption speed should be set by customer comfort, not by what the technology allows. Brands that get that sequencing backwards will win a quarter of efficiency and lose years of trust.

What advice would you give to a marketing team looking to incorporate AI into its strategy for the first time?

Don’t start with the agent. Start with a single identity key, one ID that reliably ties a customer’s activity together across every system you have, and make sure every source actually uses it. That alone resolves more of the fragmentation problem than people expect. Then sync the handful of signals that actually change a decision in real time, not your whole data warehouse, just the events where getting it wrong is expensive.

Only once those two things are solid does it make sense to let anything act with any autonomy, and even then, start with the lowest-stakes decision available, send-time or channel choice, not offer or discount logic. Autonomy comes last, and narrow before broad.


Quick Hits

A useful app or tool you’ve started using recently outside of work:

Nothing unique here: Claude and ChatGPT. I think of them as co-intelligence rather than tools. They don’t just help me write; they help me think. I use them to brainstorm, to push back on an idea, and to take a half-formed thought and see where else it could go. The best conversations leave me with a sharper version of what I was trying to say and sometimes with an idea I didn’t know I had.

Book, podcast, or resource you recommend:

The Before the Coffee Gets Cold series by Toshikazu Kawaguchi. In a small café in Tokyo, customers get the chance to travel back in time, but only until their coffee gets cold, and nothing they do can change the present. Each story is about regret, love, and the people we wish we could meet once more.

The books are short, gentle, and stay with you long after you finish them. They remind me that our past doesn’t need changing, only understanding.

Something you want to learn or get better at:

Two things. The first is Claude Code. I have spent my life building software companies, but it has been decades since I wrote code myself. AI is changing that. I have set up my own development environment, and I want to get good enough to take an idea to a working prototype on my own.

The second is writing short stories. I have been a lifelong reader of them, especially mysteries, and I still almost never guess the twist. I have written thousands of essays, but fiction is a different craft, where every word has to earn its place. Someday, I would like to write a few of my own.


For a bit more check out this interview with PCQuest:

Subscribe to the Almost Daily Update

We never share your info, we only share ours

Read More

the latest

Explore

advertisement

advertisement

Featured spotlight

Subscribe to our Newsletter

We never share your info. We only share ours.