The $0.78 AI Employee: A Sneak Peek at NJDMC 2026

Dan Van Tran is the Chief Technology Officer of Collectors, the parent company of PSA, and he’s taking the stage at NJDMC 2026 in Holmdel, NJ, on October 15. His session, “Agentic AI, Explained: How Machines Went From Answering to Acting,” is built for marketers rather than engineers: a distinction that matters, since most agentic AI content published this year has been written by and for technologists.

Here’s a taste of what he’ll cover and why it’s worth a seat in the room.

The story he’s opening with

Van Tran plans to start his talk with a specific example: an AI system he assigned a data engineering ticket for, the kind of job that would normally take a person two to three days. The system finished in 40 minutes for $0.78 in compute.

He calls it “the $0.78 employee,” and it sets up the question the rest of his session is built to answer: what does agentic AI actually mean for a team that isn’t full of engineers, and where should that team start? He’ll walk through the full story, including what the experiment revealed about where agents still need guardrails—live at NJDMC.

What is agentic AI, and how is it different from generative AI?

The short version, per Van Tran: generative AI answers a request. Agentic AI acts on it. A generative tool will draft an email if you ask it to. An agentic system can decide an email needs to be sent, write it, and send it, checking in with a human only at points it’s been told to.

He’ll spend part of his session tracing how the industry got from one to the other, roughly 70 years of AI history compressed into about ten minutes, and why that history changes how marketers should read the current hype cycle.

How does this play out in a real, high-stakes business?

Van Tran won’t be speaking in the abstract. Collectors is a trust business; people send PSA cards, coins, and games worth serious money, and the PSA’s grade on that item is the product. Deploying agentic workflows in that environment means mistakes are expensive, which is exactly why it’s a useful case study. Collectors holds a U.S. patent (12,400,462) on machine-learning counterfeit detection, and an AI-powered card scanner is now part of the PSA app.

He’s touched on the operational side of this before, including at the OverdriveAI Summit, on adding an estimated $3 billion of value in 14 months, as his engineering org grew from 22 to more than 220 people. Expect a version of that story reframed specifically for a marketing audience at NJDMC.

Where should a small marketing team even start?

This is the part of the talk aimed squarely at teams without a 220-person engineering org behind them. Van Tran will share his answer to a question he gets constantly: what’s a safe, simple first agentic AI experiment for a five-person team or a solo marketer? The short version involves picking one repetitive, well-defined task, but the specific framework is one he’s saving for the stage.

What happens to marketing skills as agents take on more of the work?

Van Tran has argued elsewhere, including in an eCornell keynote on the AI talent shift, that as agents absorb more repetitive execution, different skills start to matter more. At NJDMC, he’ll apply that thinking to marketing teams specifically, which is a conversation worth being in the room for if you’re thinking about how your own role, or your team’s structure, holds up over the next few years.

Who is Dan Van Tran?

Van Tran has spent his career at the intersection of trust and technology: a decade at FactSet, helping scale Flatiron Health through its acquisition by Roche, founding his own company, Zenauction, and now leading engineering at Collectors. Off the clock, he runs a personal AI chief-of-staff that manages his morning reports and email summaries and holds one of the world’s largest collections of Alex Morgan cards. Whether AI has changed how he collects is one we’re hoping to ask him live.

Want the full framework, not just the preview? Get your seat for NJDMC 2026: Connected Experiences — October 15, Bell Works, Holmdel, NJ.