Transparent by design.
We built VibeWorks to give builders an honest read on how they actually work with AI. That only works if you can see exactly what we do with your data, how we measure, and what research we built on. No marketing language. Just the specifics.
The three pillars
01 · Your data
We don't store your raw conversations.
Exactly what happens to your ChatGPT or Claude export from upload to expiry, filtered, sampled, and deletable in one click.
Read the data lifecycle →02 · How we measure
Five dimensions, scored honestly.
What each score means, how to read the decimal scale, how the biweekly progression works, and where the methodology is still imperfect.
Read the methodology →03 · The research
Tied to published work.
Every dimension maps to peer-reviewed research on how people use LLMs well. Here's the full mapping and how we calibrated.
Read the research →Quick answers
Do you store my full chat history?
No. We sample a filtered ~90 KB corpus for re-analysis and auto-expire it after 90 days. Full details →
What does my score actually measure?
Five behavioral dimensions scored 0.0 – 10.0 with explicit bands. See the rubric →
Can I delete my data?
Yes, one click from Settings. Scores stay, underlying text is purged. See the controls →
Is this based on real research?
Each dimension maps to a specific finding in the public literature. See the mapping →
What's the deeper technical posture?
Encryption, hashed IPs, ZIP-bomb protection, rate limits. See the security page →
I'm evaluating this for my company.
Enterprise posture and SOC 2 roadmap available under NDA. Contact us →
Ready to see your profile?
Upload a Claude or ChatGPT export and we'll score you across the five dimensions. About two minutes. Delete everything afterward with one click.
Run my analysis →