Parallel’s Parag Agrawal: Building a New Web for AI Agents

By Sequoia Capital

Categories: VC, Startup

Summary

Parallel is rebuilding search from scratch for AI agents, treating human click data as a bug and replacing it with agent feedback—a 1000x shift that makes competing with Google tractable for the first time because models are now good enough to compress information and rate relevance without human ratings.

Key Takeaways

  1. Human click data is fundamentally flawed feedback for agent-based search; agent feedback from models is cheaper, scalable, and more accurate—solving the ratings data problem that previously locked out Google competitors.
  2. Agents will do search 1000x more than humans, requiring completely reinvented technology and business models, not just incremental improvements to existing search infrastructure.
  3. The web crawl and indexing problem is expensive but no longer insurmountable for startups; the real breakthrough is using large language models as rating systems instead of hiring human raters at scale.
  4. Founders must unlearn lessons from post-product-market-fit, scaled businesses; building for agents requires betting on customers that don't exist yet and learning their needs weekly, not optimizing for hundreds of millions of current users.
  5. Web search is fundamentally a billion-to-billion matching problem (hundreds of billions of pages vs. queries); agents running this at scale will expose massive inefficiencies in how information is indexed and ranked today.

Related topics

Transcript Excerpt

Our view at parallel is that human click data is a bug and agent doing work with search should rely on agent feedback not human feedback. We believe that these models are really good at compressing information and we can benefit from a lot of the research that have gone into building models and apply it to search indexing and ranking. And so you can now make many many arguments and that's the arguments we made back then that actually now it's way more tractable as a problem because of the existence of agents not just as in technology but as a distinct customer Parog, thank you so much for joining us today. Uh we're delighted to have you on the show. For those who don't know, Parog of uh Twitter CEO fame was was CEO of Twitter um before selling it to Elon. Uh and is now back on the founder …

More from Sequoia Capital