Over the past few years, short-sellers like Culper Research and Capstone have pulled back the curtain with reports that read like financial true crime novels. I’ve raised my own questions here at ADOTAT.

But now? I’m getting anonymous emails that hit differently.

These aren’t your run-of-the-mill disgruntled employee rants. They’re specific, detailed, and unnervingly confident—the kind of messages that make you pause mid-scroll and start asking who isn’t talking publicly yet.

✉️ “It’s public info—Company 1 in the Culper report is Zeta. Kubient did it once, Zeta built a business on it. I can share exact details. We’re talking tens of millions per month.”

That’s what one anonymous insider told me. Calmly. Repeatedly. With the precision of someone who knew where the bodies were buried—and could draw the map blindfolded.

David Steinberg is not a man who whispers. He speaks like someone trying to out-run the doubts in his rearview mirror: fast, enthusiastic, and very, very rehearsed. A veteran of the dot-com boom, the telecom bust, and the slow-motion implosion of InPhonic, he’s now positioned himself as the torchbearer of AI-powered marketing.

Under his leadership, Zeta has gone public, gobbled up companies like LiveIntent, and rolled out a marketing cloud built on data, machine learning, and chutzpah.

He’s also very good in a room. Whether you’re a client, an investor, or a journalist trying to sneak in a skeptical question, Steinberg has a way of pivoting from charm to conviction faster than your AI assistant can mispronounce your name.

Their latest pitch? A shiny new framework called Zeta Answers—an AI-powered system of "multi-agent workflows" that allegedly reduces the distance between data, insight, and execution. You feed in marketing questions (“Who’s most likely to churn?”), and out comes not just an answer—but a set of automated actions: email campaigns, ad retargeting, CTV placements, and maybe even a cheerful nudge to your CRM team.

Sounds impressive. Feels futuristic. But is it real AI? Or just clever branding on top of if/then workflows and deterministic matching?

We’re told these agents operate with "autonomy." That they’re replacing dashboards with decisions. And to be fair, Zeta is investing heavily—with partnerships including NVIDIA and deep R&D into applied AI systems. The company claims that over 245 million U.S. consumer profiles are being analyzed in real time across thousands of attributes. The promise? That this intelligence makes marketing cheaper, smarter, and faster.

So why are so many industry insiders side-eyeing this rocket ship?

Let’s get specific. The most persistent concerns surrounding Zeta Global fall into three buckets:

Short-sellers have repeatedly pointed to Zeta’s pre-IPO relationship with Kubient, a now-defunct ad tech firm that was accused of round-tripping revenue—a tactic where Company A "sells" services to Company B, and then B buys it right back, making it look like real revenue passed through.

In its infamous 2021 report, Culper Research directly linked Zeta (as "Company 1") to this alleged scheme. Zeta has publicly denied booking revenue from Kubient. But the anonymous source who contacted me alleges this was not a one-time arrangement, but rather “tens of millions per month” built into the business model. If this is true—and again, if—it raises the specter of financial engineering, not operational excellence.

Zeta’s Data Cloud claims to hold 550+ million global identities, including over 245 million Americans. Impressive… until you start asking where that data comes from, how recent it is, and what protections are actually in place.

Zeta says it strips personally identifiable information (PII) and replaces it with Zeta IDs, claiming to operate within all privacy regulations like GDPR and CCPA. Critics aren’t so sure. They argue that the data is largely recycled from past acquisitions, with questionable opt-in provenance, and that "deterministic identity" may be more optimistic branding than actual compliance practice.

Zeta’s shopping spree is well-documented: LiveIntent, Temnos, Disqus, Boomtrain, and more. These assets span ad tech, email infrastructure, data enrichment, and personalization tech. But are they actually integrated into a single, seamless marketing cloud?

Zeta insists yes. Critics—and former employees—suggest that behind the scenes, much of the stack remains fragmented, and that the seamless front-end might be masking a Frankenstein’s monster of old code, clashing schemas, and duct-tape workflows.

In fairness, Zeta hasn’t been hiding under a desk. Their PR team responds. Their CFO is front and center on earnings calls. Their rebuttal to short-sellers?

15 pages long.

Key points they’d make:

And you know what? Some of that may be true. Possibly even most of it. But if there's a fire under all that AI smoke, it's going to show eventually. That’s what we’re here to find out.

This is not about dunking on Zeta for sport. It’s about untangling ambition from artifice, and differentiating innovation from inflation.

Over the next four parts, we’ll explore:

Zeta Global isn’t just chasing the AI wave — it’s trying to own the surfboard factory. With Zeta Answers, its new AI Agent Studio, and a buzzword buffet of "agentic workflows," the company is betting big on a future where marketing runs itself. No more bloated teams juggling platforms.

Just plug in a few digital agents, set your KPIs, and let the algorithms do the dirty work.

It’s a bold, ambitious vision. And let’s be clear: Zeta isn’t bluffing about its intentions. They’re putting real money behind it. In 2024, their AI-driven “consumption revenue” jumped over 40%, and the company now serves over 40% of the Fortune 100.

That’s not PowerPoint bravado — that’s market penetration at enterprise scale.

But ambition comes with baggage. Beneath Zeta’s sleek pitch lies a more complicated question: 🧩 Are we looking at the future of marketing… or a very slick demo built on stitched-together code and legacy deals?

Let’s unpack the promise, the power, and the pain points.

Zeta’s new “Agentic Workflows” allow marketers to build and deploy AI agents that can:

It’s not just automation. Zeta claims these agents are adaptive, self-optimizing, and able to act autonomously based on streaming behavioral data. And unlike some vendors still stuck in "if-this-then-that" logic, Zeta is claiming true orchestration, where one AI agent’s output informs another's decision-making in real time.

Sounds great. But even insiders acknowledge the user experience today is still a work in progress. Multiple sources describe it as “powerful but patchy,” particularly when agents rely on legacy systems built through acquisitions.

Zeta’s marketing cloud isn’t homegrown from scratch. Instead, it’s the result of over a dozen acquisitions spanning everything from identity graphs (LiveIntent) to content engines (Temnos) to ad tech (Disqus).

This kind of roll-up strategy can deliver scale fast, but it also creates what some users describe as a CRM Frankenstein:

Zeta says it integrates everything under a unified Zeta ID that allows deterministic targeting and personalized campaigns at scale. But some industry analysts argue the seams still show — particularly when it comes to which functions are powered by true generative AI vs. traditional automation.

Zeta claims to have 245 million+ opted-in U.S. identities, matched through email, mobile IDs, and behavioral data. But how opt-in is “opt-in”? Critics argue that much of Zeta’s data sourcing relies on "consent farms" — websites that offer low-value content (like giveaways or quizzes) to collect emails under questionable disclosure.

Zeta strongly denies this. In a recent interview, CEO David Steinberg even admitted he had to look up what “consent farm” meant when the accusation surfaced. Still, internal sources and bankruptcy filings from partners like Kubient raise concerns about how far back this practice may go — and whether some early client wins were underwritten by recycled PII.

Despite the smoke, here’s where things get impressive: Zeta’s monetization model is working. Not just for PR, but for Wall Street. Here's the breakdown:

Zeta’s land-and-expand approach — starting small, scaling based on ROI — has turned more than a few cautious enterprise clients into $2M+/year power users. It’s not vaporware. It’s repeatable revenue.

Let’s be fair. No platform at this scale is going to be perfect. But the core challenges facing Zeta include:

You could argue that Zeta is really building two machines:

That’s not necessarily sinister. That’s capitalism. But when the pitch is this ambitious, and the numbers this bold, transparency becomes part of the product.

Zeta Global is not your average martech company. It’s swinging for the fences with a bet that AI can finally close the gap between insight and action. And it might just be right. But behind the curtain, the platform still shows its age, its patchwork origin story, and its operational complexity.

The question isn’t whether Zeta knows where the money is. It does. The question is whether the engine driving that growth is as autonomous and intelligent as advertised — or if human intervention is still doing more heavy lifting than they’d like to admit.

If you're just skimming the headlines, you’re missing the real Zeta story—the one buried under AI buzzwords, shiny earnings decks, and conveniently timed acquisitions. This ADOTAT+ issue goes under the hood of Zeta Global, exposing how its $250M LiveIntent deal wasn’t visionary, but a well-dressed plug for a crumbling cookie-free future.

We dissect the actual guts of their AI stack (spoiler: Boomtrain redux), map M&A patterns to earnings calls like a forensic accountant with a grudge, and unpack SEC risk like it’s a game of Jenga—with Steinberg’s charisma as the top block.

Don’t settle for surface-level spin. This is where the story gets complicated—and actually useful.