AI Summary
5 min readParag Agrawal, former CEO of Twitter, is now building Parallel, a company that rethinks web search from the ground up for AI agents rather than humans. The core insight is that agents will eventually perform "a thousand X more" web searches than humans ever did, and the infrastructure built for human browsing—keyword search, click-based ranking, ad-supported economics—is fundamentally mismatched for how agents consume information. Parallel is building both the search technology and the business models to support what Agrawal calls a "parallel web" designed for agentic consumption.
Why Agent Search Is Different from Human Search
Human search is shaped by laziness and impatience. Users type short, often typo-ridden queries and click on fast-loading pages even if those pages are less authoritative. This created an entire SEO industry that optimizes for human behavior—producing pages that load quickly and put key information "above the fold" even if they are derivative. Agrawal calls this "pre-AI human slop," but notes it was a rational response to human incentives.
Continue reading the full summary in the app — free to try.
Read Full Summary →Free • No credit card required
Never miss an episode of Training Data
Get every new episode summarized in your inbox — free, ~5 minutes to read.
No spam. Unsubscribe anytime.
What you'll learn
- 1 Timestamped Outline
- 2 (00:00) **Why Parallel Exists** - The core thesis: human click data is a bug for agent search; agent feedback is the future
- 3 (01:24) **What Parallel Builds** - A technology stack allowing agents to search and use the web, with new business models to match
- 4 (02:21) **Unlearning Twitter Lessons** - Why running a post-PMF giant differs from building for a customer that doesn't yet exist
- 5 (03:48) **The Web Search Problem Defined** - A clear explanation of crawling, indexing, and ranking as a billion-to-billion matching problem
- 6 (05:35) **Why a Startup Can Compete with Giants** - Three years ago, the barriers were feedback data and infrastructure cost; agents change both
- 7 (07:52) **The Search Agent Pivot** - Launching a search agent product first, not a search engine, to trade crawl for inference-time compute
+ Full timestamped outline available in the app
Guests on this episode
Show Notes
Parag Agrawal is making a bet that goes against two decades of web search: agents will query the web a thousand times more than humans ever have, and the infrastructure built around human clicks is wrong for them. The former Twitter CEO, now founder and CEO of Parallel Web Systems, explains why Parallel treats human click data as a bug and trains on agent feedback instead. He unpacks the counterintuitive choice to ship a search agent before a search engine, building an index incrementally, and how the new Turbo product cut agentic search to 200 milliseconds. But the problem Parag keeps returning to is economic: the ad-supported internet collapses when agents show up instead of people. His fix draws on Shapley values to pay content owners for the value their pages provide agents, with real dollars reaching publishers, he predicts, within 12 to 24 months.
Hosted by Sonya Huang and Andrew Reed, Sequoia Capital
More from this podcast
Training Data →