Podcast Summary: How I Invest with David Weisburd

Episode 210: How Startups Can Avoid Being Disrupted by OpenAI w/ Eric Olson
Date: September 8, 2025
Guest: Eric Olson, CEO & Co-founder of Consensus
Host: David Weisburd

Overview

This episode explores how startups—especially those in AI—can defend themselves against disruption from colossal players like OpenAI, Meta, and Google. Eric Olson, CEO of Consensus (an AI-powered search engine for academic and scientific research), discusses the practical realities of vertical vs. horizontal AI products, the importance of focus and speed in startups, and the real nature of competitive moats in today’s AI ecosystem. The conversation also touches on talent wars, building for a hyperscale market, product development strategies, and how asset allocators might approach investing in the AI space.

Key Discussion Points & Insights

1. The Threat of AI Giants & The Horizontal vs. Vertical Debate

• The Central Question: How can startups avoid being steamrolled by generalist AI giants? (00:00)
• Olson observes a growing concern about competition from OpenAI and similar companies.
• Quote: “If a product that is as simple as a meeting transcriber in your Zoom window can't be fully disrupted by these horizontal models, there's so much room for vertical products to live.” (00:47, Eric Olson)
• Example: OpenAI's meeting transcription hasn't meaningfully disrupted more focused products like Granola.
• Finding: Startups with deep specialization (vertical products) can survive and thrive even as horizontal tools become more powerful.

2. Consensus: Mission, Market, and Growth

• Background: Consensus is an AI-powered search engine for academic & scientific research, aiming to be the next-gen Google Scholar or PubMed. (03:29)
• Target users: Students, academic researchers, faculty, clinicians, and R&D professionals across industries.
• Market Size Estimate: ~500 million potential end users globally (academic, healthcare, industry). Google Scholar sees ~50M monthly active users, PubMed about 20M. (04:29)
• Product Differentiation:
  • Olson notes Google Scholar is “frozen in time” and hasn’t evolved in 20 years. Consensus adds synthesized summaries, visualizations, in-line citations, and workflow features. (06:01)
  • Quote: “We're super lucky to be talking about a product that is pretty frozen in time. So anything we do beyond a list of links is differentiating between Google Scholar.” (08:25, Eric Olson)

3. Talent Wars in AI

• The "AI talent wars" are unprecedented. Some professionals have allegedly turned down offers of $1B over 4 years from big AI labs—more than any sports contract. (01:52)
• Consensus doesn’t compete for the rarest “frontier model research” talent, but rather hires strong classic software engineers with AI skills, a more sustainable and accessible talent pool for vertical applications. (02:06)
• Quote: “We're still competing against big companies for folks like that, but it's more of like what startups have always had to do…” (02:53, Eric Olson)

4. What Makes Vertical AI Products Defensible?

• Delight & Focus: Products like Granola succeed because they obsessively focus on a specific user problem, adding layers of detail and care that horizontal platforms don’t prioritize. (11:41)
• Nat Friedman’s Analogy: “I could hire somebody off the street to clean my house and they would probably do just about as good of a job as somebody who has a cleaning service. But I still go to the cleaning service. Cause they might just be 10% better...and I'm willing to pay a little bit extra for that...” (12:30, paraphrased, Nat Friedman via Eric Olson)
• Org Inertia: Even at big companies, without company-wide focus, side projects don’t receive the same inertia and excellence. Startups have an edge by concentrating all resources on one problem. (14:00)
• Risk Tolerance: Startups can also take on more product risk than FAANG-scale companies, who face massive consequences for errors. (15:19)

5. Building Amid Hyperscale Change

• Urgency & Speed: In a market where AI advances monthly, startups must have "complete urgency" and work at maximum pace to win mindshare before models become good enough for their use cases. (17:15)
• Anticipating Model Advances: Sometimes, Consensus ships with more expensive or limited models, betting that new, cheaper, better AI models will emerge shortly—speed outweighs waiting for perfect conditions. (18:15)
• Quote: “Speed is just... more important than it's ever been before in startups today with, with the moving markets.” (18:19, Eric Olson)

6. Product Development: Customer vs. Intuition

• Balance: About 70-80% of product development is driven by direct user requests; 20-30% comes from internal insight, intuition, and strategic bets. (19:34)
• Pain over Features: It’s vital to focus more on customers’ pain points than on building exactly what they ask for. (20:45)
• Quote: “It's a classic like user interview. Best practice is like always take a little with a grain of salt. The question is usually never, what do you want us to build? It's more like what are you feeling when you use the product and what are you trying to solve for what is your problem?” (20:54, Eric Olson)

7. On Moats and Defensibility in AI Startups

• The Myth of the Startup Moat: Unless you have brand, distribution, or true tech breakthrough (like Nvidia’s hardware lead), most software startups lack traditional “moats.” (21:28)
• Startup advantage: relentless focus and speed, not unique, inimitable assets.
• “Moat” is often a VC trope—Olson calls it a “reddish yellow flag” if asked as a superficial checkbox. (23:04)
• Quote: “Moats are kind of a myth...Your quote unquote moat as a startup is your focus...and your speed and ability to innovate and take risks.” (21:28, Eric Olson)

8. Investor View: How to Play the AI Space

• Stack Exposure: Olson recommends asset allocators invest across the whole AI stack (hardware, infrastructure, apps), not just one layer. (24:32)
• Old Rules Apply: Stick to timeless fundamentals: back the best founders tackling important problems—AI doesn’t change the “ABCs of investing.” (26:30)
• Beware overheated mega-rounds, and invest within the constraints of your allocation style and philosophy.

Notable Quotes & Memorable Moments

• On Vertical Startups vs. Giants:
  “There's never not going to be a market for [vertical specialization] if you do that incredibly well. Even if intuitively...some products should be swallowed up by a model…people generally overestimate how much that will happen.”
  (09:56, Eric Olson)

• On Focus as Startup Advantage:
  “Your moat against big players is your focus. And getting into every nook and cranny of your problem… Even the most capitalized, smartest people… can only truly be great at a finite number of things.”
  (10:40, Eric Olson)

• On Risk Tolerance:
  “If they put an LLM into Google Scholar and it summarized the paper that said vaccines cause autism...they might lose $50 billion of market cap. Is that risk worth it for them?”
  (15:30, Eric Olson)

• On Product Evolution:
  “Google Scholar...is kind of like a fun way to see what Google used to look like. It's the same interface, list of blue links... no summary, no great interactability.”
  (06:01, Eric Olson)

Timestamps for Key Segments

• State of the Market & Vertical vs. Horizontal Defense — 00:00–01:02, 08:42–11:27, 13:17–15:56
• Consensus Company Update & TAM — 01:14–04:29
• Google Scholar vs. Consensus — 05:52–08:42
• AI Talent Wars — 01:52–03:23
• Focus, Delight & Moats — 11:41–15:56, 21:20–23:15
• Building in an AI Hyperscale World — 16:52–19:18
• Customer vs. Intuition in Product — 19:18–21:20
• Investor Perspective on AI — 23:56–26:54
• Closing Thoughts & How to Follow Consensus — 27:10–27:40

Conclusion

Eric Olson argues that startups can remain resilient and even thrive in the world of omnipresent AI giants by relentlessly focusing on user problems, moving with urgency, and leveraging the inherent organizational advantages of narrow focus and higher risk tolerance. The moats for startups, he suggests, are not technological secrets but rather the fast, fanatical execution and patient, user-centric iteration that large companies can rarely match. For investors, the fundamental rules of software investing remain unchanged: back the best people solving real problems—the same advice that has worked for decades.