AI Agents Are Writing Web Scrapers Now | Full Meetup Recording #1 https://www.youtube.com/watch?v=7CukuGZ25gs?utm_campaign=blog-posts&utm_activity=ORS&utm_medium=social&utm_source=mastodon
AI agents can now write, run, and self-heal your web scrapers, and ...
"A work of art."
That's how Proxyway described the Byteful platform in their 2026 Proxy Service Awards, where they named us Newcomer of the Year. It's an independent benchmark, so not our own claim.
Read more: https://proxyway.com/research/proxy-service-awards-2026
New Jersey Globe: Nearly 400 local newspapers sue OpenAI, Microsoft over alleged copyright theft. “The massive coalition of local newspaper publishers filed a federal lawsuit today against OpenAI and Microsoft, alleging the technology companies systematically copied copyrighted reporting from nearly 400 local newspapers to train and develop commercial artificial intelligence products, including […]
https://rbfirehose.com/2026/06/25/new-jersey-globe-nearly-400-local-newspapers-sue-openai-microsoft-over-alleged-copyright-theft/
New Jersey Globe: Nearly 400 local newspapers sue OpenAI, Microsoft over alleged copyright theft. “The massive coalition of local newspaper publishers filed a federal lawsuit today against Op…
The country with the most active residential IPs in our network is the United States, by a wide margin.
Top 7 by live active IPs: US 967K, UK 302K, Canada 234K, China 136K, Brazil 106K, Italy 95K, Germany 91K (Byteful dashboard, June 2026).
Why it matters: a 35M global pool with 94K IPs in Italy behaves nothing like one with 940K. The shape of the coverage matters more than the headline number, so ask for active counts in your target country.
I run GLM 5.2 inside Claude Code with the same tools, the same skills, and the same agent loop. Three environment variables is all it takes. https://www.zyte.com/blog/how-to-run-any-model-inside-claude-code?utm_campaign=blog-posts&utm_activity=ORS&utm_medium=social&utm_source=mastodon
Research and Development at a Web Scraping company https://www.youtube.com/watch?v=VYfLRRVx2tI?utm_campaign=blog-posts&utm_activity=ORS&utm_medium=social&utm_source=mastodon
An interview with Zyte's head of R&D. Read it here: https://www.zyt...
More instruction, worse output. Zyte's head of R&D on why telling your agent exactly what to do can blind it to the obvious answer. https://www.zyte.com/blog/the-best-agent-skill-is-the-one-that-says-the-least?utm_campaign=blog-posts&utm_activity=ORS&utm_medium=social&utm_source=mastodon
For a human, 10ms is invisible. For an AI agent chaining thousands of requests, it's everything.
So we traced one residential request end to end: closest of 12 POPs for TLS (~15ms), pick from 35M+ exits (under 8ms), session (~40ms), target round trip (~300ms).
410ms median, down from 690ms. Most of it is physics, not us.
Is GLM-5.2 really closing the gap to Anthropic - and at just a fraction of the cost - or is it just more AI hype? I think so, and let me show you why. https://www.zyte.com/blog/why-im-adding-glm-5-2-to-my-agentic-coding-arsenal?utm_campaign=blog-posts&utm_activity=ORS&utm_medium=social&utm_source=mastodon
For data gatherers, negotiating anti-bot tech is not for the faint-hearted. That’s why we built a multi-layered system that treads lightly but brings a host of tactics to the table. https://www.zyte.com/blog/how-we-built-a-320000-strategy-web-access-hero?utm_campaign=blog-posts&utm_activity=ORS&utm_medium=social&utm_source=mastodon
Zyte
Byteful
ResearchBuzz: Firehose
