Field notes

Building this in public.

Notes from the research and decisions behind MagFlow AI — what we learned from Clay's own users, what things actually cost to run, and why some obvious-sounding features didn't make v1.

Why your outbound data gets worse the further you are from San Francisco

Clay's data marketplace is genuinely strong — for companies that raised a Series A and got tech press. Everyone else is working with thinner coverage than the marketing suggests. Here's what we found reading through dozens of real user threads.

Why we cut video and PDF, and built a corroboration engine instead

MagFlow AI started with plans for personalized video and PDF deliverables, matching feature-for-feature against Clay. Here's why both got cut, and what the effort went into instead.