Podcast Summary

Behind the Numbers: an EMARKETER Podcast

Episode: AI-Driven Media Management, with Gigi and Amazon Ads (Part 2)

Date: December 23, 2025
Host: Jeremy Goldman (Senior Director of Content, EMARKETER)
Guest: Adam Epstein (Co-Founder & CEO, Gigi)

Episode Overview

This episode is the second installment of a special miniseries focusing on how AI is revolutionizing media management, particularly in partnership with Amazon Ads. Jeremy Goldman interviews Adam Epstein, CEO of Gigi, for a deep dive into vertical AI agents, innovation in advertising technology, and predictions for the future of AI-driven media buying.

Key Discussion Points & Insights

1. Industry Recognition & Amazon Partner Ecosystem

• Award Recognition: Gigi recently won the Global Innovation Technology Innovation Partner Award at Amazon Unboxed.

  • Adam remarks on Amazon's unique, supportive approach to partnerships, distinguishing them from other big tech platforms.
  • “Amazon has very rightly identified that the best way to achieve the optimal experiences for their advertisers and customers is through partners…a really harmonious partnership.” (Adam, 02:28)

• Partnering with a Rapidly Evolving Platform:

  • Amazon continues to release features and level up its ad products.
  • Amazon’s API-first infrastructure, influenced by AWS, lets partners access functionalities early, fostering innovation among technology providers.

2. AI’s Transformative Role in Media Management

• Difference in AI Approach:

  • Amazon’s agentic AI (Ads Agent) “raises the floor” by making DSP tools more accessible for new advertisers.
  • Gigi’s AI “raises the ceiling,” giving sophisticated advertisers bespoke, deeply tailored tools leveraging agentic AI for unique strategies (04:34).
  • “Within agentic AI, we no longer buy software, we hire software, we train and we manage it just like we would a team member.” (Adam, 04:59)

• AI as a Tailwind:

  • Amazon’s push into agentic AI validates the importance of an AI strategy for all, but sophisticated advertisers demand more customization than baseline tools can offer.

3. Defining “Vertical AI Agents”

• Clarification & Buyer Heuristics:
  • Vertical AI agents aren’t just generic tools—they replace or augment specific human roles, such as media managers.
  • True vertical AI is judged by whether it reduces necessary headcount for specialized functions (07:06).
  • Token-based business models can be a litmus test: if pricing isn’t tied to token usage, likely little real AI is in play (09:17).

4. Core Building Blocks of Vertical AI Agents

• Three Major Components:

  • 1. LLM Models:
    • Early reliance on a single model (GPT-5), now moving to a ‘constellation’ of frontier and fine-tuned models specific to media.
  • 2. RAG (Retrieval-Augmented Generation):
    • Ingests proprietary documents and process docs to make each AI agent ‘unique’ to the customer; not just generic LLM intelligence (10:36).
  • 3. Tool Calling & MCP Servers:
    • AI must interact with APIs (“tools”) for actionability; Amazon’s MCP server will soon let the agent flexibly choose tools for various client needs, streamlining integration (11:45).

• Strict Data Segregation:

  • “Are you taking all this data and best practices and laddering it up to a master model? …the answer is emphatically no.” (Adam, 13:02)
  • Ensures privacy, legal compliance, and bespoke customer solutions.

5. Guidance for Non-Technical Leaders Building with AI

• Curiosity over Credentials:
  • Adam emphasizes the value of “insatiable curiosity” and a learning culture over technical backgrounds.
  • GG’s success comes from a willingness to dive deep, even from a non-engineering start.
  • “Whenever we hire a new team member, we like to say like, you’re going to get a black belt in AI by becoming a member of the GG team.” (14:33)

6. Looking to 2026: The Next Wave in AI-Powered Media

• Gigi’s Roadmap:

  • Expansion beyond Amazon DSP to other DSPs and retail media platforms is the next focus (16:22).

• The “Deck Creation” Challenge:

  • Automated client deck generation remains unsolved but is a top priority as generative AI matures.

  • Adam predicts near-future breakthroughs as multimodal models (e.g., Google’s image generation) improve, soon allowing human-level, auto-generated client presentations (17:20).

  • “Once the technology of deck creation, image creation, becomes commoditized…vertical AI agent[s] can uniquely…build the best possible deck for their verticals.” (18:40)

• Broader Industry Trends:

  • Rapid commoditization of advanced tasks—what’s hard today will quickly become table stakes.
  • “AI might not actually be there yet, but it’s moving so incredibly fast that you need that vision, and then the AI will probably get there at some point in the near future.” (Jeremy, 19:03)

Notable Quotes & Memorable Moments

• On Partnership with Amazon:

  • “It’s the exact opposite with Amazon…it’s allowed for companies like Gigi and many other technology companies to deliver a lot of enterprise value and delight to customers.” — Adam (02:17)

• On the Agentic AI Model:

  • “The ads agent is raising the floor…we’re doing the exact opposite. We’re raising the ceiling on the most sophisticated advertisers.” — Adam (04:42)

• On What Makes AI “Real”:

  • “If the agent you’re hiring is able to enhance that role that requires you to hire less of those people…then that’s a true vertical AI agent.” — Adam (07:39)

• On Learning AI from Scratch:

  • “We just have this insatiable curiosity to learn.” — Adam (14:40)

• On AI’s Future in Creative Tasks:

  • “We have all of this advertiser context…how can we automatically create a deck based on the template…? The technology wasn’t there. But now…I think we’re getting really close to having automatic deck creation.” — Adam (17:39)

Timestamps for Key Segments

• [01:29] Award win and Amazon partner ecosystem
• [03:43] AI strategy comparison: Amazon Ads Agent vs. Gigi
• [06:20] What is a vertical AI agent?
• [09:17] How to evaluate true AI in products
• [10:36] Core building blocks: Models, RAG, Tool Calling
• [11:43] MCP server explained
• [13:14] Data isolation and enterprise customization
• [14:03] Adam’s career journey; advice for leaders
• [16:17] Gigi’s 2026 roadmap: multi-platform expansion
• [17:20] The holy grail: AI-powered deck creation
• [18:53] Reflections on tech limitations, rapid AI evolution

Final Takeaways

• AI is fundamentally changing media management, not just by improving efficiency for beginners (“raising the floor”), but by enabling entirely bespoke, sophisticated workflows for seasoned advertisers (“raising the ceiling”).
• True vertical AI agents must fully integrate into workflows, replacing or enhancing real human roles—not just providing shiny dashboards.
• The next leaps in AI for media will likely come from multi-platform integrations and the achievement of automated, high-quality creative outputs.
• Curiosity and adaptability are the keys for organizations and leaders to succeed with AI—not necessarily deep technical experience.

For listeners, this episode offered an actionable, inside look at where AI-driven media management stands and where it’s heading, from a company shaping its evolution in real time.