Podcast Summary:

AdTechGod Pod – Ep. 97: "Agents, Not People: How AI is Rewriting Ad Operations" with Joe Hirsch

Host: AdTechGod
Guest: Joe Hirsch (CEO, Swivel; Former CEO, SpringServe)
Date: September 16, 2025

Overview

In this episode, AdTechGod hosts Joe Hirsch, a recognized leader in ad tech with deep roots in CTV (Connected TV), OTT, and ad serving. The conversation explores Joe's career journey, his motivation for founding Swivel, and how AI-driven automation—"agents, not people"—is transforming the role of ad operations professionals. The episode also touches on industry shifts, the evolving skills ad ops teams need, and the future of human-machine collaboration.

Guest Background and Career Journey

Joe Hirsch's Entry into AdTech

• Started in junior high with affiliate marketing via early search engines, promoting Amazon’s affiliate program (then offering 15% revshare)—primarily selling cheat code books for Super Nintendo. (02:28 – 03:20)
• Launched first network (Haguenet Media) with partner Hagen Major, representing 100+ video game and wrestling sites.
• Sold network (renamed Click Hype) during college, joined Spark Studios, then founded performance house Yellowhammer Media Group.
• At Yellowhammer, saw a need for better video ad serving, prompting co-founding of SpringServe.
• Led SpringServe to acquisition by Magnite after eight years.

"That's kind of a brief history of the universe." – Joe Hirsch (06:19)

The Problem: Repetitive Labor in Ad Ops

Manual Scaling’s Limits

• Ad servers aren’t "broken," but the business model is: companies require a growing number of ad ops people for growth, with most work being highly repetitive. (06:44 – 08:41)

"You require an infinitely scaling number of ad ops people to address a scaling business... Those people are, they're all doing the same thing." – Joe Hirsch (07:15)

The "Swivel Chair" Problem

• "Swivel Chair": The inefficient practice of operating across many browser tabs and platforms to execute ad ops tasks. (09:01 – 09:40)

The Swivel Solution: AI-Powered Operations

Automating Tedious Workflows

• Swivel is designed to automate operational workflows across platforms—from insertion orders and campaign trafficking to in-flight optimization, yield management, and post-campaign reporting. (09:41 – 12:53)
• Enables 24/7 operation, scaling actions far beyond human capacity.

Real-World Ad Ops Workflow (as it exists today)

1. Trafficking insertion orders into ad servers—often partially automated, requires manual completion.
2. Adding creative elements, audience targeting, inventory selection—typically operator tasks.
3. Ongoing campaign optimization and reporting, based on human-generated routines.

Swivel’s Approach

• Aims to automate these steps, freeing ad ops to focus on strategic initiatives, not rote mechanical tasks.

AI Adoption: Replicate, Scale, Improve

Swivel’s Three-Phase Approach

(13:23 – 17:45)

1. Replicate Human Playbooks:
   • Documents existing "playbooks" for troubleshooting (e.g., how to respond to underpacing).
   • Automates sequential steps a human would take—like adjusting pacing or targeting rules.

   "The first thing that you might do is change even daily pacing to daily ASAP... and if that doesn't resolve delivery, you might change the frequency cap..." – Joe Hirsch (14:02)

2. Scale the Process:
   • Once automated, increases the action frequency—hourly or even intra-hourly, not just daily or weekly.
3. Improve on Human Decisions:
   • Eventually, AI suggests or implements optimizations humans may not have considered.

Example:

• Human ad ops might analyze app-level conversion data, exclude low-performing inventory, and adjust targeting. Swivel automates this entire process, running it at much higher frequency and across more objects.

"You can only do them when you're awake... Swivel allows you to do it at scale, across all objects, 24 hours a day at unlimited capacity." – Joe Hirsch (11:54)

Tech Details: Data Ingestion & Real-Time Action

Scaling Ad Ops with AI

(17:45 – 20:06)

• Humans can manage ~1,000–2,000 campaign actions per day; Swivel can multiply that by 10–100x.
• Real-time or near-real-time decisioning constrained only by how quickly systems can provide fresh data (every 15 minutes, for instance).

"A team might be able to produce a thousand or 1,500 or 2,000 actions in a day. And Swivel can take that number and 10x it... 15,000, 20,000, even 100,000 actions per day." – Joe Hirsch (19:33)

The Future: AI Agents as Platform Users

THE NEXT AD OPS USER: NOT A HUMAN

(20:34 – 23:11)

• Soon, the primary user of ad platforms will be agents (AI)—not people.
• Analogous to how web traffic is increasingly made up of bots, crawlers, and agents rather than direct human interaction.

Human Oversight Still Critical

• Precision is "paramount" in ad ops; errors are costly.
• AI augments, but does not fully replace, human decision-making—humans "consent" to actions at all stages.

"In ad ops, precision is paramount. An ad ops error is a costly mistake... Keep a human in the loop at all points such that a human is consenting to the scaled workflow." – Joe Hirsch (21:55)

Industry Impact, Human Role, and the Need to Adapt

Impact on Ad Ops Workforce

(23:11 – 24:41)

• AI won’t "replace" all ad ops, but will reduce the need for teams of 25+, empowering smaller, more expert teams.
• Operators must adopt and grow with AI, or risk being left behind.

"If you're an operator and you ignore the trend, I think that's more dangerous than adopting it and becoming an expert in it." – Joe Hirsch (24:05)

Manual Work is on Borrowed Time

• Manual tasks like data entry, trafficking, and routine optimizations are fading; learning to manage and harness AI tooling is vital.

Notable Quotes

• On the mission of Swivel:

  "I think about those workflows as broken more so than the ad servers themselves." – Joe Hirsch (08:26)

• On the shift to AI agents:

  "The future of ad ops is that an agent will be the primary user of the ad platform." – Joe Hirsch (20:40)

• On integrating AI in ad ops:

  "AI is critical. AI can improve human behavior, but it doesn't mean the human is no longer involved." – Joe Hirsch (22:24)

• On professional advice for ad ops teams:

  "I would encourage people to adopt these technologies and learn how to use them. Because unless everyone's wrong, these things aren't just going to disappear overnight." – Joe Hirsch (24:12)

Memorable Moments & Timestamps

| Topic | Speaker | Timestamp | |--------------------------------------------------------------|--------------|-----------| | Joe’s "brief history of the universe" | Joe Hirsch | 06:19 | | The "Swivel Chair" problem explanation | Joe Hirsch | 09:09 | | Automating a human’s campaign optimization process | Joe Hirsch | 13:23–17:45| | Ad ops action scaling: “Swivel can take that number and 10x it”| Joe Hirsch | 19:33 | | Agents, not people, as the future ad platform users | Joe Hirsch | 20:34 | | "Ad ops error is a costly mistake...human in the loop..." | Joe Hirsch | 21:55 |

Conclusion

This episode underscores a pivotal shift in the advertising technology industry: the rise of AI-powered agents as operational workhorses. Joe Hirsch advocates for embracing automation—not for the sake of replacing people, but to elevate the value and capacity of ad ops teams. The future belongs to those willing to learn, adapt, and harness these powerful tools, ensuring humans remain at the helm of strategy while agents handle the "swivel chair" grind.