Voices of Search Podcast

Episode: How Brands Can Prepare for the Direct-to-Agent Era of AI Discovery
Host: Tyson Stockton
Guest: Brian Stempak, Co-Founder & CEO, Evertune
Date: December 15, 2025

Episode Overview

This episode explores how the rise of AI-powered agents (like ChatGPT, Gemini, and others) is radically transforming how consumers discover, research, and purchase products online. Host Tyson Stockton and guest Brian Stempak discuss actionable strategies for brands to remain visible and influential in this emerging “direct-to-agent” environment, where decisions are increasingly made by AI on behalf of users, not just by users themselves. The conversation offers key insights into generative engine optimization, data-driven content strategies, and the evolving nature of search, purchase, and brand influence in the age of AI.

Key Discussion Points & Insights

1. The Shift from Search to AI-Powered Discovery

• Explosion of AI Prompts: ChatGPT processes over 2.5 billion prompts daily—330 million from the US alone. This isn’t just replacing search; it’s the dawn of a new discovery paradigm.

  “They’re the start of a new discovery model where agents are making decisions for users. Optimizing for end consumers isn’t enough anymore.” (Tyson Stockton, 00:43)

• Changing Consumer Behavior:

  • Traditional search: 3-4 word short queries; quick sessions with click-driven outcomes.
  • AI discovery: 20-25 word prompts; users give more context, expect comprehensive—often multi-minute—conversations and answers.
  • Shift from multi-step, multi-site research to consolidated, agent-managed investigation and even direct purchasing within the AI tool.

  “On average, people are spending six or seven minutes in AI... The consideration cycle or purchase funnel is shortening, it’s condensing.” (Brian Stempak, 02:06–03:45)

2. Complexities of Monitoring AI Visibility

• Probabilistic Responses: Unlike static search rankings, AI models generate varied answers each time, depending on prompt history, user data, and model version.
• Data-Driven Approach: Evertune aggregates millions of prompts to build a detailed, geo/persona/model-based “heat map” of brand recommendations across AI agents.

  "We prompt on average a million times a month per brand to understand what’s happening in the world of coffee makers, in the world of Keurigs…” (Brian Stempak, 04:58)

3. Breaking Down the Models: Variability, Data Sources, and Output

• Significant Variance Across Models & Modes:
  • ChatGPT core model vs. model with live search plugin can swing results by 20-30 percentage points for the same query.
  • Each model (ChatGPT, Gemini, Google, Claude, Llama) weighs sources, data, and brands differently.
  • Licensing deals (e.g., Google’s with Reddit) significantly affect which sources are prioritized.

  “Even on ChatGPT versus ChatGPT plus search, there can be meaningful differences... These are driven by what data sets models have access to.” (Brian Stempak, 09:18)

4. Strategic Implications for Brands

• Long-Term vs. Short-Term Optimization:

  • Core Model Influence: Slow, long-lasting—impacts retrain cycles and enhances organic awareness.
  • Search Citation Influence: Faster, more tactical—target citations, PR, affiliate, or sponsor content on high-authority, high-citation third-party sites.

  “The models don’t retrain every single day... But when you do influence that model, it’s longer lasting... On the search side, that’s more similar to traditional SEO where you can see faster changes.” (Brian Stempak, 12:57)

• Actionable Recommendations:

  • Create highly targeted, authoritative content on brand sites and strategic external platforms.
  • Identify and prioritize “hotspot” third-party sources frequently cited by AI agents in your space.

  “Five to ten content pieces can have a meaningful impact if you’re talking about the right things... If you really focus on what subject areas you’re trying to move the model on.” (Brian Stempak, 15:14)

5. The Direct-to-Agent Era: What Happens When the AI Agent Buys for You?

• Entry of AI in Purchase Decisions: ChatGPT, for example, allows direct purchasing (affiliate commission model) within the AI experience, initiating a new era in e-commerce.
• Rise of Trusted AI Agents: For recurring or low-consideration purchases, consumers might outsource the entire decision—brand influence switches from the consumer’s mind to the AI’s algorithms and data understanding.

  “Imagine a world where I’m saying, okay, AI grocery shopping agent, you know that we're drinking Keurig at home... Find me the best price for... and go order me 200 of them... That’s where this gets really interesting, where the agent... decides.” (Brian Stempak, 16:39)

6. Content Strategy Transformation for AI Agents

• The New Content Principle:

  • For AI, exhaustively document every detail of your brand, products, and features—agents “want” as much information as possible.
  • For B2B: Make RFP-level details public and accessible; AI agents use this to “draft” recommendations.

  “An AI model has all the time in the world to learn about your brand... Step one is making sure the AI model knows everything about your product set... And that’s not always how marketers think.” (Brian Stempak, 23:23)

• From Coverage to Specialization:

  • Build content that covers key use cases, buyer personas, and specific decision criteria.

  “It means creating content that maybe you wouldn’t have otherwise... building your content around [strategic use cases] versus trying to own everything.” (Brian Stempak, 27:04)

7. Quality vs. Quantity: Risks and Rewards

• Flooding the Zone?
  • Current AI models reward more comprehensive, open data, but a shift toward evaluating trust and source quality is expected—flimsy content might lose ground as models mature.

  “If Keurig just published a thousand articles saying here’s the reasons Keurig is the best... over time, the model is biased towards quality. We already see that... in the sources they cite.” (Brian Stempak, 29:49)

8. Third-Party Data and Content Source Insights

• Key Data and Content Partners:

  • AI models favor sources they have licensed or judge as authoritative: Reddit, YouTube transcripts, Travel + Leisure, Wirecutter, category review sites.

  “When they have deemed it enough quality to pay for it, they cite it more often... Another common theme, we see YouTube quite a bit, a lot of the YouTube transcripts.” (Brian Stempak, 31:54)

• User-Generated and Independent Content: Gaining ground again due to objectivity and authenticity (Reddit is “difficult to game”; up- and downvotes reflect community sentiment).

  “It’s vetted by people... even if you try to promote your brand on Reddit, you’re going to get called out for it... That’s why you see some of the models paying so much money to license data from Reddit.” (Brian Stempak, 33:39)

• Niche and B2B Opportunities:

  • AI falls short in less-documented or paywalled verticals, sometimes citing unlikely sources (e.g., job postings). Opening up proprietary content can create first-mover advantage.

  “If the good information might sit behind a paywall... in some cases we’re encouraging: you need to make that more open.” (Brian Stempak, 34:24)

9. International & Multilingual Dynamics

• Global AI Model Gaps:

  • In non-US markets, models may over-index on US brands and data due to training set biases; nuances and localization are still evolving.

  “Even for a user in Australia asking questions about their local homeowners insurance companies, it’ll sometimes bias towards American answers.” (Brian Stempak, 37:12)

• Legacy and Confusion Issues:

  • Large, historic brands often have an advantage; models sometimes confuse recent brand changes or mergers unless corrected by new data or searches.

  “The core models sometimes get confused by rebranding... the models can get confused by that, they get confused by brand hierarchies.” (Brian Stempak, 38:27)

Notable Quotes & Memorable Moments

• On the AI Agent becoming the decision-maker:

  “The consumer has said: I don’t actually care. I’m going to let the AI bot decide. And so that’s a really powerful shift.” (Brian Stempak, 17:48)

• On rethinking marketing for AI:

  “For this audience, it’s: the more data the better.” (Brian Stempak, 25:53)

• On leveraging niche content:

  “If that’s [a job posting] their best set of data, it’s a pretty thin answer. And if you’re a brand in that category, that’s an opportunity... That’s a flag that hey, this is a place where the models don’t know very much. It’s an opportunity to inform them.” (Brian Stempak, 34:24)

Timestamps for Key Segments

• 00:43 – AI discovery vs. traditional search: user behavior, prompt length, and impact on brand interaction
• 02:06 – Evertune’s approach: panelist data, consumer research, and large-scale AI prompt analysis
• 06:25 – Measuring brand visibility in volatile, multi-agent AI market
• 09:18 – Variability between models: ChatGPT, Gemini, Llama, and the impact of data licensing
• 12:57 – Influence strategies: long-term model retraining vs. short-term citation/PR wins
• 16:39 – The rise of AI-powered commerce and agent-led decision making
• 23:23 – Content and product detail for AI: B2C vs. B2B best practices
• 27:04 – Shifting focus: from broad coverage to persona/use-case-driven content
• 31:54 – Third-party sources and publisher partnerships’ growing importance
• 34:24 – Content gaps = competitive opportunity (esp. for B2B and niche verticals)
• 37:12 – International insights: model regionalization and legacy brand effects

Takeaways for Brands

• AI discovery is not just “search 2.0”—it is a fundamentally new, agent-driven model where influencing the AI (not just the person) is key.
• Brands should shift content strategies to maximize detail, depth, and availability of product, pricing, and category information—especially for AI to access and use.
• Prioritize content partnerships, PR, and citations on platforms that AI models trust and license (Reddit, YouTube, Wirecutter, etc.).
• Monitor model output globally, track persona/geography/model variances, and leverage uncovered content gaps for strategic advantage.
• Accept that marketing for AI is a “coverage play”—the more you feed models (including use case and technical details), the more “discoverable” you become.

For SEO, content, and digital marketing leaders: Rethink your traditional search approach now. The era of agent-driven discovery is here, and mastering generative engine optimization is crucial to dominating tomorrow’s organic landscape.