build Featured How We Think About Pricing Compute for a Scraping Tool Why we price a scraping tool on compute, not seats: usage-based credits that track browser time, proxy tier and AI calls, plus a real free trial.
build Featured The Proxy-Tier Ladder: How a Scrape Escalates When It Gets Blocked A block is a signal to climb, not retry. How a scrape escalates one proxy rung at a time, and the re-probe that downgraded 24 of 52 pinned tiers.
build Featured We Gave Every MCP Tool a readOnlyHint: Agent-Safety by Design Trawl's MCP server gives nine tools to AI agents, and two of them spend compute. We hand-annotated all nine so a grounding agent never fires a billable scrape run by accident.
build Featured What Actually Breaks a Scraper in Production A column of zeros looks like one bug. It's four: nav fails, selector rot, blocks, geo incoherence, each needing the opposite fix. Diagnose before you heal.
build Featured Why We Run Scrape Scheduling on Mongo, Not Kafka Why we coordinate scheduled scrapes on MongoDB instead of a message broker: the incident that forced it, one atomic lock, and the honest trade-offs.
News Featured Trawl Beta Launch, Plus MCP Upgrades The June 2026 beta launch and MCP upgrades: a sharper MCP server for AI agents, fairer billing, and reliability polish. Trawl is now open, free plan, no credit card.
