The founder and MD of the Customer engagement platform believes that agentic AI could help solve marketing's improbable problems.
Marketing technology was supposed to make marketing measurable. However, fractured stacks and poorly developed use cases defeat the very purpose it was acquired for.
In a new paper called Rewiring Martech: From Cost Centre to Growth Engine, McKinsey and Company found that most marketing leaders were unable to quantify returns on martech spend. In fact, not one of the 50 or so senior marketing leaders at Fortune 500 companies that the consulting major interviewed could clearly articulate the ROI of their organisation's investment in martech and how it's driving value.
Many senior marketers have often complained that they lose customers because of a problem with their martech stack. Disconnected systems make it difficult to share data, leading to inaccurate personalisation or missed opportunities.
Currently, brands are spending 90% of their budgets on adtech (to acquire new customers). But according to Rajesh Jain, founder and managing director of Netcore Cloud, what's most troubling is that 70% of this budget is spent on customers that the brand already acquired, but who have become dormant. This means that the brand is spending money to monetise the same customer over and over again because martech has left the door open for customers to be lost.
Jain argues that this is where agentic marketing comes in. Earlier this year, Netcore Cloud unveiled nine agentic AI agents, each specialising in a core marketing function from journey orchestration to content generation to analytics.
Marketers were already using generative AI to generate content and communications faster, better and cheaper. But the way Jain envisions agentic is in helping marketers solve impossible problems, such as reactivating dormant customer bases, reducing customer acquisition costs, or achieving true hyper-personalisation beyond basic demographic segments.
"What these single-agent and multi-agent systems will do is solve the most important problem, which is how do I get N-equals-to-1 personalisation," he said.
Identifying the sweet spot
The marketing problem is effectively very simple -- how do I get the right message to the right person at the right time, Jain said. However, because segmentation has been huge, coarse, not enough data or lacked the ability to process all the data, consumers are often inundated with irrelevant offers.
With agentic, this process is streamlined, Jain says. Brands can now create up to 500 segments and push out tailored, contextual content in minutes.
"In cases of unforeseen events like cyclones or airplane crashes (life insurance companies often see an uptick in demand in the immediate aftermath) or timely moments such as a Taylor Swift concert, brands need to be able to react very quickly. That's where very fine customer segmentation and brand twins come in," he told Campaign.
Brand twins are digital replicas of customers that continuously learn from every interaction, building increasingly accurate models of individual preferences and behaviours. As Jain explains it, this agent-twin ecosystem enables a 'Department of One for a Segment of One' -- where small marketing teams can deliver truly individualised experiences to millions of customers simultaneously.
Speed unlocks new opportunities, but also new mistakes. Jain said there is human oversight at every stage of the segmentation and content creation process to minimise this.
The user has final approval over the wording of the problem statement/brief, how segments are defined, how channels are defined or used, the content created and finally, how this is disseminated. At every stage, they can intervene or edit according to brand guidelines.
"Of course, if you're creating 500 pieces of content, humans can't validate all of the content. But there are very good guardrails in place to ensure that the systems don't 'hallucinate'," he shared.
One of the guardrails is to create a brand 'sandbox' for the AI to operate with, such as defining brand colours, tone of voice, and brand guidelines. "This essentially eliminates a lot of distractions and deviations and makes AI more 'predictable' in the output and scheme," he shared.
Jain envisions a day when the system has learnt the brand codes to such an extent that human intervention is no longer needed, but he says, we're not there yet.
One of the fears with AI is what happens when the system is down. The global AWS outage in October 2025 is one example. Ecommerce and banking apps were unable to process orders, while gamers were locked out of online gaming platforms.
Brushing off these instances, Jain maintained that while outages like this are often widely reported, they rarely affect more than 0.01% of operations. "My guess is that we were still at 99.99% uptime. These things do happen, and of course, we have SLAs in place, but these outages are generally under that 0.01%," he said.
The icing on the cake
One of the reasons why martech failed is that it is treated as a one-time purchase and not a growth engine. The above-mentioned McKinsey report found that rather than tying outcomes to revenue growth, customer lifetime value, or strategic business goals, most marketing teams simply track operational metrics such as email sends and open rates, or campaign delivery metrics such as impressions delivered and reach.
Netcore wants to nip that problem in the bud with agentic. To ensure a return on investment, the company is experimenting with a variable pricing model, which it calls Alpha. Instead of charging fixed platform fees plus per-user or per-message costs, the platform now offers variable pricing tied to actual revenue outcomes. Under this pricing model, it aligns its targets with client marketing teams. If both sides exceed expectations, it shares in the 'alpha', the upside achieved beyond projected targets.
"Throughout its history, martech has struggled because it's primarily input-based. We typically price on three metrics: number os messages sent, monthly transacting users and in the case of a search business, API calls. These are all input-driven. No one in this industry has done outcome-based pricing," Jain explained. These pre-agreed outcomes vary by industry.
.jpg%26h%3D630%26w%3D1200%26q%3D75%26v%3D20250320%26c%3D1&w=1280&q=75){kind=small}
