Oct 24, 2026 · 60 min
AI agents could force a new web infrastructure stack
Ex-Twitter CEO Says We’re Underestimating AI Agents by 10,000x
If agents generate vastly more web activity than people, search, identity, security, and publisher economics may all need redesigning.
- 1AI agents may generate thousands of times more web activity than humans, overwhelming infrastructure built around human browsing.
- 2Parallel is building agent infrastructure spanning search, browsers, sandboxes, action APIs, and connectors to private data.
- 3Proactive agents could shift the internet from users pulling information toward systems continuously delivering relevant information and actions.
Don't miss
Agrawal describes the web’s possible transition from users pulling information through searches to proactive agents pushing information and actions to them.
The brief
Former Twitter CEO Parag Agrawal says AI agents could generate 5,000–10,000 times more web activity than humans, a scale that demands new infrastructure rather than simply more conventional search.
His company, Parallel Web Systems, is assembling that stack: search, browsers, sandboxes, action APIs, and connections to personal or company data for agents that can do more than answer questions.
The conversation moves from GPU scarcity and scaling a team from 30 to 120 employees to the harder question of access: how websites distinguish helpful agents from malicious bots.
Prompt injection and agent identity expose the central tension. Systems must act proactively across the web while defending against content designed to manipulate their decisions.
Agrawal’s larger forecast is a shift from a pull web, where people request information, to a proactive web where agents continuously surface needs and take action.
Weekly email
State of the Pod, every Monday
The week's most-discussed names and stories from podcasts, with the moments behind them. Free, five minutes, unsubscribe in one click.