By Venk Chandran
My First Nvidia purchase was the NV1 in 1995. It was about $400 CAD (this was a lot for me at the time) and had a whopping 2MB of RAM. It was compatible with the SoundBlaster audio card, and I was ready to rip on some games and use the new CD-ROM Encyclopedia. Reader: The games did not work well on my Frankenstein build of a computer running a Pentium chip; nothing did.
At the time, there were two ways to get your game working, and you could divide us into two camps: those of us who believed in the power of prayer - that the game would load and the disc would hum instead of spin wildly. Others with more agency (impatience?) just continually reinstalled DOS until it worked.
And when it worked, I remember thinking this was it. We’ve reached the pinnacle of modern technology. I can play Panzer Dragoon (SEGA). Of course, it turns out GPUs can… do more? Who knew?
Ok, well Jensen Huang did. Last week was my first time attending the NVIDIA GTC Conference. And Jensen said something during his keynote that’s been rattling around my head. Agentic AI has hit an inflection point. Not a future inflection point. Now.
He wasn’t talking about better chatbots. He was describing a new class of systems: ones that reason, plan, take action, and operate continuously over long horizons. “An AI that could generate became an AI that could reason,” he said. “An AI that could reason became an AI that could do work.” Adobe, Cisco, CrowdStrike, Salesforce, SAP - dozens of the world’s largest enterprise software companies were named as partners building agents on NVIDIA’s platform. The message was hard to miss: AI isn’t just a feature anymore. It’s becoming the operating layer of the enterprise.
I was thinking about what these grand proclamations mean for B2B marketing during GTC. And then, over the past few weeks, I’ve met with a few dozen CMOs and marketing leaders and got a much more complicated answer.
What’s Actually Happening on the Ground
The conversations I’ve been having with marketing leaders don’t sound like keynote optimism. They sound like urgency - and sometimes alarm.
A marketing technology leader at a large enterprise technology company put it plainly in a recent conversation: “We are seeing real impact and drop of traffic in our web experiences - this last six months or so.” They weren’t talking about a mild dip. They were describing a structural shift as more buyers begin their research inside AI systems rather than typing queries into search bars and clicking through to company websites.
A CMO at a fast-growing software company articulated the newer version of the problem: “More of their buyers’ pre-funnel education is happening in their own session with their own AI.” Her buyers are forming opinions and shortlists before they ever touch her marketing. By the time they arrive - if they arrive - they’re already halfway through a decision someone else helped them make.
A VP of global demand generation, someone who has been thinking about this longer than most, framed it in terms of infrastructure: “None of us are marketing in silos anymore.” Her concern isn’t just about individual campaigns - it’s about how marketing data, intent signals, and content intelligence flow across systems, teams, and now, agents.
Every one of these leaders is dealing with the same underlying reality: headcount is shrinking, budgets are tightening, and yet the expectation to drive pipeline hasn’t budged. The workflows they’ve built over the last decade - ABM campaigns, demand-gen plays, sales enablement - were designed around human teams operating fragmented tools. Marketo, Demandbase, Salesforce, content platforms - the institutional knowledge of how to string these together lives in the heads of a handful of people. When those people leave (or their roles are eliminated), the machine stops.
But the more striking thing these leaders are telling me isn’t about their current team. It’s about who they expect to be engaging with their content next year.
Agents are becoming B2B buyers.
Not metaphorically. Literally. AI agents are now doing research, evaluating vendors, consuming content, and informing purchase decisions alongside - and sometimes instead of - human buyers. And when those agents go looking for information about a product category, they aren’t clicking through your marketing experience. They’re the most ruthless buyers imaginable: scanning APIs, reading structured data, bypassing brand experiences entirely. They just want the answer.
Most B2B marketing stacks were built entirely for humans. And that’s becoming a problem.
Content Was Always About Context
At PathFactory, we’ve spent a decade on a single premise: B2B buyers use content to self-educate, and if you understand what they’re engaging with, you can build better experiences, surface better context for sellers, and close more deals.
That premise hasn’t changed. But the world around it has changed dramatically.
Content Intelligence - knowing what content resonates with what buyers - has to evolve into Context Intelligence. Because content is context. For a human buyer reading a whitepaper, for a seller about to make a call, and now, for an AI agent evaluating a vendor landscape, the underlying need is the same: reliable, structured, accurate signal about what buyers care about and where they are in their journey.
At a fireside chat in San Francisco in February, I heard Aravind Srinivas - CEO of Perplexity - make an observation that stuck with me: “horizontal AI will ultimately beat verticalized software”. The models are getting better faster than any single-domain application can compete with. And Perplexity is proof of concept - they operate their own go-to-market with just very few enterprise sellers, using AI so deeply for campaigns, research, and customer engagement that they don’t need a traditional commercial structure to run a growing enterprise business.
That’s not an anomaly. It’s a preview.
When I look at what GTC announced - the NVIDIA Agent Toolkit, OpenClaw, which Jensen called “the most popular open-source project in the history of humanity” (for the record mine is called Jean-Clawd Van Damme), the entire ecosystem of partners building autonomous enterprise workflows - and I map that onto what these CMOs are telling me about how their buying committees are changing, one thing becomes very clear:
The B2B marketing platforms of the next decade won’t be apps that marketers log into. They’ll be context layers that agents and humans both consume.
What We’re Building
At PathFactory, we’re in the middle of that transformation. Here’s how I think about it:
We’re shifting from a SaaS application to a Context-as-a-Service platform.
That means the buyer journey data we capture - what accounts engaged with what content, at what stage, through what channel - stops being an output for a dashboard and becomes a queryable, programmable substrate. A context graph that any agent or application in the GTM stack can tap into.
We’re also rethinking what the B2B buyer experience actually means in an agentic world. One of the most important things we’ve been building is what we call Vibe Sessions: a persistent memory of the buyer built from everything they’ve done - the content they’ve consumed, the questions they’ve asked our agent, the signals they’ve emitted across every interaction. The goal is to capture that memory as close to real time as possible, so that at any given moment we understand the “vibe” of the buyer: where they are in their thinking, what they care about, how ready they are to move. Not a form fill. Not a lead score. A living, continuously updated read on the B2B buyer as they actually engage. That changes everything about what you serve them next - and it’s the kind of context that both human marketers and agents need to do their best work.
On the agent side, the shift we’re making is more fundamental than adding AI to existing workflows. Most B2B marketing applications - including ours - were built as thick UX layers: wizards, step-by-step builders, configuration screens designed for humans to click through. That model made sense when humans were the only operators. It doesn’t scale anymore.
We’re moving from an application layer to an orchestration layer. Instead of a marketer navigating the full complexity of enterprise ABM configuration - integrations, branding, audience segmentation, content mapping - an agent takes a natural language brief and does the work, while the marketer stays in the loop as editor and decision-maker, not operator. The institutional knowledge of your best ABM practitioner gets encoded into the agent as skills. The APIs become its tools. What used to demand significant time and coordination across tools and teams can now happen in an afternoon.
One of the reasons our agent experiences perform the way they do is an infrastructure choice we made early: we run ChatFactory on Groq, which delivers significantly faster response times and inference performance than standard GPU setups. That matters in a B2B buyer conversation - latency kills trust.
It was gratifying to see NVIDIA validate that bet at GTC, where they unveiled the Groq 3 chip as a centrepiece of their Vera Rubin platform following their $20 billion acquisition. The inference performance race is real, and we’re already building on the winning side of it.
We’ve already seen early proof that this kind of context-aware, high-performance engagement is a different category entirely: one customer reported that 95% of buyers who engaged through our ChatFactory agent and converted ultimately became marketing-qualified leads. That’s not a chatbot stat. That’s what happens when the experience is genuinely intelligent.
And we’re building MCP connectivity - making PathFactory’s intent data and engagement signals directly queryable by other agents in the stack. Because if agents are going to do work on behalf of buyers and marketers alike, they need to be able to read reliable context. That context layer is what we’re building.
This is what Jensen was describing. Not AI as a feature. AI as the operating layer. PathFactory’s job is to be the trusted context layer that the operating layer runs on.
What CMOs Are Beginning to Ask
The CMOs who get this aren’t asking “how do I use AI to write more content faster?” They’re asking a harder question: “When my buyers - human and agent - show up looking for context about my category, am I even findable? And when they find me, is what they find reliable enough to act on?”
That’s the right question. And it’s the one we’re building toward.
One marketing leader, in the same conversation where they described the traffic drop, asked something that’s been on my mind ever since: “Do we even need ABM platforms anymore?” They weren’t dismissing them - they were asking whether the platforms needed to fundamentally reinvent themselves.
The answer, I think, is yes. The ones that survive will be the ones that become infrastructure for the agentic layer, not just interfaces for human marketers.
But the more immediate challenge I hear from CMOs isn’t philosophical. It’s operational: how do I actually scale my team with agents? Because the pressure is real - smaller budgets, fewer headcount approvals, but the same pipeline targets. The instinct is to bolt AI onto existing workflows. That almost never works. Agents don’t make broken processes faster. They expose them.
Scaling with agents requires three things most marketing organizations haven’t yet done. First, your content and data have to be structured and accessible - agents can’t operate on siloed, untagged, poorly governed assets. Second, your institutional knowledge has to be externalized. The ABM playbook that lives in your best marketer’s head needs to be encoded somewhere an agent can actually use. And third, your stack has to be composable - tools that speak to each other through APIs and shared context, not a patchwork of disconnected point solutions where every handoff is manual.
That’s a meaningful amount of work. But it’s the foundation. Without it, “scaling with AI” just means more noise at higher speed.
The future of B2B marketing isn’t larger teams with more tools. It’s smaller, more focused teams with clearly encoded institutional knowledge, paired with agents that do the orchestration work on top of a reliable context layer. The marketers who thrive will be the ones who shift from operators to editors - setting strategy, exercising taste, and letting agents handle the execution at scale.
We’re not there yet.
But back in 1995, I was convinced I’d reached the pinnacle of modern technology. I could play Panzer Dragoon. “That Lightning Crashes” by this new band LIVE was not going to be just another one hit wonder. I was wrong then.
The marketers who figure out agentic infrastructure in the next 18 months may not be at the pinnacle either - but they’ll be the ones actually playing the game while everyone else is still waiting for it to load.
Venk Chandran is CPO at PathFactory. He writes about AI, product strategy, and the future of B2B go-to-market.