# The Marketing AI Pulse Brief for March 2026: Agents, Shrinking Moats, and the Rise of Trust!

By Aby Varma · Published 2026-03-26 · Updated 2026-11-03 · Professional AV on MarketScale
Canonical: https://www.marketscale.com/industries/pro-av/the-marketing-ai-pulse-brief-for-march-2026-agents-shrinking-moats-and-the-rise-of-trust
Creator hub: The Marketing AI SparkCast

> The Marketing AI SparkCast brings you the March 2026 Marketing AI Pulse Brief, with host Aby Varma of Spark Novus and co-host Matt Cyr of Loop AI. They cover how agentic AI is moving marketing from creating content to executing work, why true adoption still lags, and why trust and brand visibility now decide who wins, with stories on NVIDIA, FedEx, Google, Adobe, Figma and Grammarly.

## Key points

- Agentic AI is moving marketing from content creation to executing actual work
- Adoption of AI marketing tools still lags despite technological advancement
- Trust and AI search visibility are shifting what marketers can control

The Marketing AI SparkCast brings you the Marketing AI Pulse Monthly Brief, a curated look at the most relevant stories shaping the field and what they mean for CMOs and marketing leaders. In the March 2026 episode, host Aby Varma of Spark Novus and co-host Matt Cyr of Loop AI work through five stories that point the same way: AI in marketing is moving from creating content to executing work, and trust is becoming the asset that decides who benefits.

## Agents move from chatting to doing

Aby opens the episode with NVIDIA, where founder and CEO Jensen Huang introduced two layers for enterprise AI agents: a build layer for designing and creating agents, and a run layer for operating them inside a business. Today, Aby notes, there is no common way to build agents. Microsoft, Google, Salesforce, Adobe and SAP each have their own methodology.

The build layer, NVIDIA's agent toolkit, is an open platform with three elements: Nemotron reasoning models that let agents work through multistep tasks such as campaign planning and optimization, OpenShell guardrails for privacy and compliance, and AI-Q for running agents faster and more cost-effectively at scale. The run layer, NemoClaw, builds on OpenClaw and turns it into an enterprise-ready platform that coordinates agents across systems, manages workflows and adds the security controls OpenClaw was missing. Aby points to the ecosystem as the real signal: Adobe, Salesforce, SAP, Cisco and ServiceNow are working with NVIDIA to bring these capabilities into platforms marketers already use.

> For marketing leaders, that shift is clear to me, where the question is not just what AI can create, it's what AI can actually run. — Aby Varma, The Marketing AI SparkCast

Matt, who is building a platform for marketing agencies built on agentic structure, says the weight of the announcement comes from its source. A company with NVIDIA's clout is telling the market that agents are not a passing fad. He also sees the era of tools built as LLM wrappers around ChatGPT coming to an end, and a next wave centered on agentic behavior and massive infrastructure.

Aby's FedEx story, drawn from a Wall Street Journal article, shows the model in practice. FedEx is building an agentic system that turns business objectives into coordinated execution across functions, with marketing operations and campaign management called out specifically. A manager agent breaks a goal like launching a campaign into steps, worker agents carry out tasks such as data, messaging and channel execution, and an audit agent reviews output for accuracy, compliance and quality. For CMOs, Aby says, the job shifts from managing workflows to designing how work flows through systems, with governance embedded in the process.

On the episode, Matt raises the conversation he says marketers are not yet having, and will need to:

> What are we allowing AI agents to do on their own? What keys to which cars are we gonna give them? — Matt Cyr, The Marketing AI SparkCast

Without set rules for when an agent asks for help or sign-off, he warns, companies could get themselves into "some significant tough corners."

## Everyone is using AI. Almost no one has implemented it.

Matt brings a Supermetrics survey that cuts against the 80 to 90 percent usage figures common in industry reports. Only 6% of marketers said AI was fully implemented in their marketing operations, 80% said they feel pressure to adopt AI, and 37% said they lack a clear AI strategy from leadership. He flags that Supermetrics has a dog in the fight, and that he doesn't know anyone who is fully done with AI implementation. Still, the findings track with what both hosts see in the market: adoption as "death by a thousand cuts," with groups trying separate tools, not talking to each other and not measuring impact.

> There's value in experimentation, but I think we're moving very quickly to the point where experimentation has got to become focused implementation. — Matt Cyr, The Marketing AI SparkCast

Aby notes there is no standard definition of "fully implemented." He recalls a Fortune 100 marketing executive who said AI was in their workflows. Pressed for detail, the executive meant notetakers. Matt's prescription: find workflows AI can handle, measure the benefit, name a person or team to lead the effort, and codify governance now instead of waiting for the perfect moment. Aby adds the advice he gives clients.

> You've got to attach your AI initiatives in marketing to some sort of a business value. Otherwise, it's just an experiment that is going to eventually wither away. — Aby Varma, The Marketing AI SparkCast

## Google's AI Overviews rewrite the rules of visibility

Matt walks through a BrightEdge report finding that Google's AI Overviews are 44% more likely than ChatGPT to display negative information about a brand, which works out to about 23,000 negative responses for every 1,000,000 queries. BrightEdge's CEO suggests the cause is how Google sources its overviews: to give a more rounded answer, it pulls from old sites, historical information and negative reviews. Search, Matt says, is no longer something a brand can control 100%.

The metrics are changing with it. Time on page and total views are giving way to visibility, meaning being cited or referenced inside AI answers. Matt's advice for CMOs: respond genuinely to negative reviews, even old ones, and don't let bots do it. Remove outdated pages that no longer help your current story. And invest in subject matter expert content that helps people answer a question, solve a problem or move forward in a buying decision.

Aby cites Rand Fishkin's figure that AI desktop search in North America is still about 3%, with Google dominant. He argues brands can't ignore the shift anyway, because ignoring it is how you create "Blockbuster and Netflix type stories."

## Shrinking moats at Adobe and Figma

Aby turns to Adobe, where CEO Shantanu Narayen announced he plans to step down after eighteen years, with no successor or timing yet named. Narayen moved Adobe into a successful cloud subscription model and architected its AI direction, but Canva and Figma are rolling out generative AI that lets anyone create professional-looking content from simple prompts, and Adobe's stock has fallen. Matt adds that Figma's stock dropped 12% in a matter of days after Google released its own vibe coding product.

> For all these companies, the quote unquote moat is shrinking. — Aby Varma, The Marketing AI SparkCast

Aby's takeaway for marketers is about martech strategy. Avoid being tied too tightly to one ecosystem, because Canva, Figma and Google keep adding capabilities that leapfrog one another. Derivative content that once called for an expensive enterprise platform can now be produced at speed with templates and prompting. Martech loyalty, he says, needs to be reexamined, and teams need to get more agile.

## Grammarly and the price of broken trust

Matt closes the episode with Grammarly, which he calls the gateway drug to AI for a lot of people. In 2025, it released Expert Review, letting users summon specific, named experts to review their writing. None of the experts knew they were on the panel. None had been asked, none were compensated, and one of them was dead. The people involved found out from a story in The Verge.

Matt reads it as a symptom of the "ship first, ask questions later" ethos in AI, and an unnecessary hit over a nice-to-have feature at a time when larger platforms are adding capabilities that can do a lot of what Grammarly does. His advice for CMOs: ask whether a feature needs to be AI at all, set clear guardrails for AI in customer-facing experiences, and talk to customers before rolling anything out.

> Don't do what Grammarly did here: burn a decade's worth of trust to the ground for a nice-to-have feature. — Matt Cyr, The Marketing AI SparkCast

Watch the full March 2026 Marketing AI Pulse Brief from The Marketing AI SparkCast for the complete conversation between Aby Varma and Matt Cyr, or listen on Spotify. You can join the Marketing AI Pulse community for free at marketingaipulse.com.

## Sources

- [The Marketing AI Pulse Brief for March 2026 (Spotify)](https://open.spotify.com/episode/6IpRzFuLRSsCMcKQ3k1eK8) (The Marketing AI SparkCast)
- [The Marketing AI Pulse Brief for March 2026 (video)](https://marketscale-4.wistia.com/medias/3z8vf8bm5o) (The Marketing AI SparkCast)

Tags: Aby Varma, Matt Cyr, The Marketing AI SparkCast, Marketing AI Pulse, Spark Novus, Loop AI

---
Source: MarketScale, https://www.marketscale.com/industries/pro-av/the-marketing-ai-pulse-brief-for-march-2026-agents-shrinking-moats-and-the-rise-of-trust. Published for AI indexing and citation; cite the canonical URL. Site guide for agents: https://www.marketscale.com/llms.txt