20VC20VC: Five Predictions for a World of Agents | The Ads Business Model Will Die | Biggest Lessons from Working with Elon Musk at Twitter with Parag Agrawal, Parallel
26 Sept 2026 · 57 min · Episode page ↗The rise of AI agents necessitates a fundamental re-architecture of web search and its underlying business models, moving from human-centric ad revenue to agent-centric data monetization and event-driven information retrieval.
Summary
Parag Agrawal, former Twitter CEO and now CEO of Parallel, discusses how the proliferation of AI agents will transform the internet's infrastructure and economics. He argues that current web search paradigms, built for humans, are entirely unsuited for agents, necessitating new technical approaches and business models, particularly a shift away from advertising towards paying for valuable data and insights at an unprecedented scale.
Interesting ideas
Agent-Driven Web Search Requires a 1000x Efficiency Leap. AI agents will search the web at a rate orders of magnitude higher than humans, demanding a 10-100x efficiency improvement in compute for web search. This shift also requires diverse optimization for speed, cost, and accuracy, as agent needs vary widely from instantaneous voice responses to deep, background research, making traditional '10 blue links' irrelevant.
The Advertising Business Model is Obsolete for Agents. The traditional ad-supported internet model fails when agents, not humans, consume content, as they don't see ads. Agrawal's company is building an 'AdSense for agents' to pay content owners directly for the marginal value their data provides, ensuring incentive alignment and preventing content from being blocked from agent access. This transition could dramatically increase the overall market size for web content.
Data Transactions at Inference Time are the Next Big Market. As intelligence becomes cheaper, the ability to transact for unique, valuable insights and data at inference time will become critical. This is currently an underserved market, as exemplified by agents needing granular access to proprietary data (like PitchBook) in ways that current seat-based or browser-scraping models cannot sustainably support.
The Web is Shifting from 'Pull' to 'Push' for Agents. For persistent, always-on agents, the web will transition from a 'pull' model (agents query for information) to a 'push' model (the search system actively monitors and notifies agents of relevant changes). This event-driven approach, like an 'LLM-powered Google Alerts,' is significantly more compute-efficient and enables new functionalities for agents constantly working on a user's behalf.
Questions worth thinking about
If traditional advertising models collapse for content consumed by agents, what new economic incentives will emerge for creators of valuable, long-form content vs. ephemeral, ad-friendly content?
How will companies with significant market power navigate the choice between blocking agents (to protect existing business models) and embracing them (to capture new value), and what are the long-term competitive risks of each strategy?
Written by AI from the episode's full transcript. It can get things wrong: if something here looks off, tell us at hello@podpings.com.