Odd Lots Podcast Summary

Episode: Dmitry Shevelenko on Perplexity's Vision for Reshaping the Internet
Date: November 6, 2025
Hosts: Joe Weisenthal & Tracy Alloway
Guest: Dmitry Shevelenko, Chief Business Officer, Perplexity
Recording Location: Lazard Foursquare Conference

Episode Overview

This live Odd Lots episode features Bloomberg’s Joe Weisenthal and Tracy Alloway interviewing Dmitry Shevelenko, Chief Business Officer at Perplexity, a leading AI-powered information platform. The conversation explores how Perplexity aims to reshape the internet, the future of news media and publishing, business models in the age of AI, how trust and information quality are maintained, and the ongoing transformation of user experience on the web. The episode is interspersed with questions from the live audience.

Key Discussion Points and Insights

1. Human Questions vs. Machine Answers

• Theme: The unique human value in journalism and information
• Dmitry Shevelenko:
  • Perplexity can answer questions, but humans excel at asking them. The core of journalism is “the spark that comes from editorial judgment.”
  • Quote (03:01): “Perplexity may have the answer, but Perplexity does not have an innate desire to be curious. And that’s what you guys have. And that’s why journalism is so important... it’s about the questions, not necessarily the answers.”
• Implication: Journalism’s future lies in curiosity and context, not just producing answers.

2. The Future of the Web & Publisher Models

• Joe’s Provocative Q: Why will websites still exist in an AI answer era?
• Dmitry:
  • The editorial process is about framing and determining what’s worth asking. AI can’t replicate that. People come to AI for deeper understanding after being inspired by journalistic work.
  • Perplexity is experimenting with programs (e.g., Comet Plus) to develop new business models for publishers.

3. Trust, Transparency, and Source Quality in AI

• Tracy’s Q: How does Perplexity decide what’s reliable?
• Dmitry:
  • Trust is the scarce commodity of the future; transparency is the foundation.
  • Perplexity shows users all sources for its information, enabling their own judgement.
  • Uses algorithms to flag consensus and outliers, but the system is not perfect.
  • Quote (05:28): “The thing that we believe that’s going to be most scarce in the future is not intelligence, it’s trust... the bedrock of trust is transparency.”

4. AI Agents, Big Tech Competition, & User Alignment

• Discussion around Amazon’s pushback on Perplexity's agent (related to its Comet browser AI).
• Dmitry:
  • The “first 30 years” of the Internet featured “lopsided AI” designed to serve big tech. Perplexity’s mission is to give powerful AI to users, aligning incentives.
  • Amazon and Google structure product for their business, not the user; Perplexity seeks to reverse this.
  • Quote (08:41): “There is a big opening for an independent, neutral AI player that is aligned with users... Our success metric is, are users willing to pay for a Perplexity subscription?”

5. Differentiation from OpenAI and Google

• Joe’s Q: What makes you different from OpenAI’s ChatGPT?
• Dmitry:
  • Perplexity optimizes for accuracy above all else, not just engagement.
  • Neutrality: Aggregates the best large language models (LLMs), maintains independent search infrastructure.
  • Quote (09:54): “ChatGPT... built a product that is optimized for engagement... Part of being neutral, part of being independent, trustworthy, accurate, is using the best possible technology to answer the question in the best possible way.”

6. Business Model & Revenue Streams

• Subscriptions: Major revenue from consumer and enterprise subscriptions.
• Enterprise Traction: A small sales team (5 people) has achieved 8x business growth.
• Ads: Not interested in rebuilding click-based ad models.
• Innovative marketing: Agents could surface truly valuable offers, not just ads.
  • Quote (11:14): “We’re excited about building products that are user aligned, that also create marketing value for brands and advertisers... but this isn’t going to be rebuilding AdWords.”

7. AI Fundraising & Private Market Dynamics

• Current Environment: AI fundraising is profoundly active, with lines between rounds increasingly blurred.
• Investor Conviction: Many investors are power users of Perplexity, so they have first-hand conviction in the product’s value.
• Caution on Speculation: Rejects “SPVs” and misaligned capital—Perplexity seeks direct, long-term investors.
  • Quote (16:38): “We want investors who are getting direct access to Perplexity and who are in it for the long haul.”

8. Flashy Moves: Chrome Bid & Investment Appetite

• Discussion of Perplexity’s high-profile (but failed) $35B bid for Chrome.
• Dmitry: Serious about the bid; ready financing from major funds if Google had been forced to divest.

9. Unit Margins, User Profitability, and Cost Structure

• Profitable Users: Paying (Pro) users are profitable; free users are not, but their activity is considered R&D.
• Subscription Tiers: Added “max” $200/mo tier for power users—solving the issue of unprofitably heavy consumption.
• Biggest Expense: Compute/inference cost for answering questions, not salary or pre-training large models.
• Scalability: As usage increases, compute for inference will require ongoing investment.

10. AI in Financial Services

• Productivity Gains: Perplexity and related AI tools seen as a force multiplyer, not automation:
  • 80% time-saving on first draft work.
  • Expands the number of new initiatives employees can take on; “not about replacing investment bankers, it’s about giving them tremendous leverage.” (25:07)
• Democratizing Finance: Perplexity Finance aims to equip retail investors and non-terminal users with Bloomberg-like capabilities, broadening access.

11. Journalism in the Age of AI: Survival & New Agreements

• Media remains viable if it adapts, as the best publishers are.
• Sees publisher-platform pacts (e.g., Comet Plus) as an evolution of Apple News+, allowing for shared subscription revenue and broader collaboration.
  • Quote (29:31): “Subscriptions are a very powerful business model for media. I think there’s a lot more to do there, and we want to play a leading role in doing that.”

12. Critical Skills & the Future of Work with AI

• Most important skill: The ability to ask uniquely good questions.
• Prompting: "A good question embodies all known knowledge and pushes the frontier; not just prompt engineering."
• Advice for Media Execs: Lead with curiosity and a drive to find new forms of leverage.

13. Web & Browser Power, Distribution, and Ad/Game Resistance

• Chrome's Strength: Massive distribution but wedded to AdWords, a vulnerability as direct answers overtake link-clicking.
• Gaming/AiSEO: Perplexity aims to be resistant to “AI SEO” manipulation; analytics tools exist but are not actionable.
  • Quote (34:58): “If you can game a Perplexity answer, it means it’s not accurate and trustworthy.”

14. Personalization, Memory, and Avoiding Filter Bubbles

• Transparency: Memory features let users see and delete “learned” preferences, preventing content pigeonholing.
  • Example: Dmitry deleted “pomegranate supplements” memory from his own profile (36:43).

15. Pace of Change and the “Bubble” Question

• Rapid Paradigm Shifts: Key is to focus on what remains the same: trust and accuracy.
• Agility: “I have absolute conviction that six months from now I’m going to have a top three priority that today I don’t know what it is.” (37:43)
• Bubble?: AI will have significant productivity ROI, but some companies are overcapitalized, risking imbalances.

16. Limitations & Failures in AI Today

• Joe: Points out how many tasks still fail in current AI agents.
• Dmitry: Progress is rapid; what failed two months ago may already be solved. Users should keep testing.

Notable Quotes & Memorable Moments

• On Journalism’s Enduring Value:
  • “Perplexity may have the answer, but Perplexity does not have an innate desire to be curious… it’s about the questions, not necessarily the answers.” (Dmitry, 03:01)
• On the Bedrock of AI Trust:
  • “The thing that we believe that’s going to be most scarce in the future is not intelligence, it’s trust… the bedrock of trust is transparency.” (Dmitry, 05:28)
• On Publisher Incentives:
  • “Part of being neutral, part of being independent, trustworthy, accurate, is using the best possible technology to answer the question in the best possible way.” (Dmitry, 09:54)
• On AI’s Business Model:
  • “We’re excited about building products that are user aligned, that also create marketing value for brands and advertisers... but this isn’t going to be rebuilding AdWords.” (Dmitry, 11:14)
• On Future-Proofing:
  • “The hard thing to predict is not what’s going to change, but what’s going to remain the same.” (Dmitry, referencing Jeff Bezos, 37:43)

Timestamps for Important Segments

• [03:01] – Why journalism and human questioning still matter
• [05:28] – How Perplexity builds trust and transparency into its model
• [08:41] – The importance of user-aligned, neutral AI business models vs. big tech’s approach
• [09:54] – Technical differentiation from OpenAI's ChatGPT
• [11:14] – Perplexity’s business model, subscriptions, and approach to advertising
• [16:38] – Investment, private market dynamics, and avoiding speculative capital
• [19:00] – Perplexity’s $35B bid for Chrome: rationale and seriousness
• [20:20] – Profitability, unit margins, and cost structure of AI services
• [25:07] – The impact of AI in financial services and productivity
• [29:31] – Vision for the future “front page” of the Internet and publisher agreements
• [32:25] – Key skill sets for executives in the age of AI: the art of questioning
• [34:58] – Gaming the answer: preventing 'AI SEO'
• [36:43] – Personalization, memory, and user control over AI “learned” preferences
• [37:43] – How to run a company when technology’s evolving at breakneck speed
• [39:09] – Bubble? The risk of overcapitalization in today’s AI industry
• [40:11] – Acknowledgement of AI system limitations and rapid ongoing improvement

Conclusion

This episode presents a nuanced vision of the future of information, work, and media in an AI-first world. Dmitry Shevelenko expresses optimism for AI’s role as an enabler, not a replacer, of human ingenuity—especially in journalism and professional services—while outlining Perplexity’s business philosophy around transparency, trust, and truly user-centric services. The challenges of monetization, the shifting incentives of publishers and platforms, and the technical race with giants like Google and OpenAI are all interrogated with directness and humor, offering a rich look at the evolving internet landscape.