Agents need adblocking, too
summary
Argues that adblocking stays essential for agentic, headless browsing: Brave pushes adblocking into the AI era to reduce visual-LLM token use and speed up agent loops, scrapers already embed adblock rust to improve training data quality, and an LLM-powered Cookie Crumbler runs in production with open-sourced crawl reports to proactively fix cookie notices. Also shows Model Context Protocol (MCP) servers let agents skip pages entirely (and can be forked to strip ads), so filter logic simply moves into agent tooling.
AI-written summary · about
chapters (7)
- 0:00Introduction to agentic browsing
- 3:14Ad blocking in AI agents
- 6:23Training data cleanup with ad blockers
- 8:54Cookie Crumbler for filter lists
- 11:19Model Context Protocol explained
- 17:26Future of ad blocking and AI
- 27:19Q&A session
related talks
53mAI Didn’t Kill the Web, It Moved in!
21mPERCIVAL: Making In-Browser Perceptual Ad Blocking Practical with Deep Learning
30mAI Agents Unchained
15mThe Death of the Browser
01h 19mEffective Transparency and User Controls: Are We Close to a Breakthrough?
58mAI Agents Will Change How We Build the Web
21mBuilding Agents (the hard parts!)
40mAI Coding Agents and how to code them