Heist was built to publish Heist's content. A month in, here's what we learned by being our own worst user.
The first tool that breaks under its own weight is usually the tool you didn't use yourself. This is what snapped first — and what the team rebuilt because of it.
Breakage #1: the empty-Brain problem
A fresh signup with no Brain data got generic content. We knew this theoretically. Living it was worse. Fix: Smart Onboarding. First 10 minutes now teach the Brain aggressively — brand profile, 3 docs, 1 persona, then rate the first 10 drafts.
Breakage #2: voice drift by week 4
Four weeks in, our own content started sounding slightly corporate. Fix: Voice Drift Detection. The Brain now compares every new draft against the voice fingerprint from week 1.
Breakage #3: the preview gap
First version had a generic preview. Posts looked clean in the editor, weird on-platform. Fix: pixel-accurate previews per platform.
What we didn't break
The core architecture held. Ten layers, closed loop, one Brain per brand. Every fix above was a layer inside the existing architecture, not a pivot.
If you wouldn't use the tool yourself, you're not the product team — you're the marketing team.