Built · AI
Shipping AI people can trust
Structured prompts, scores that explain themselves, and output that reads naturally in Arabic and English.
- EN/ع
- bilingual by default
- Why?
- every score has a reason
- Live
- results as they land
- 01
Symptom
AI output that people could not trust: scores with no reasons, and Arabic that read like a translation.
- 02
Root cause
Free-form prompts produced free-form answers, so results were hard to check and inconsistent between languages.
- 03
Fix
Structured prompts with a clear rubric, scores returned with their reasons, and both languages generated as first-class output.
- 04
Impact
Results people can read, question and act on, in the language they work in.
Debug log
- $define the rubric before the prompt
- $return a reason with every score
- $generate Arabic natively, not as a translation
- $show results as each one lands
- ✓resolved
A thousand welcomes