Every operator we know defaults to Clay + Apollo + Phantombuster + Bardeen + Dripify. The pitch is “let the experts handle scraping.”
We don’t. We build all of it in-house — ix-browser for navigation, our own enrichment pipelines, our own outreach orchestrators.
Two reasons. First, the unit economics: SaaS scrapers cost $200-2000/mo per seat per agent. With a fleet, that’s $20-200k/yr just to send messages. Killed by margin. Second, the autonomy story: every external dependency is a place where the agent has to ask permission, surface an error, or wait for a human. That breaks the compounding loop.
First-party tooling is more work upfront. It’s the only thing that scales.