AI audit vs rubric — AutoDS
An independent Workers AI LLM scored AutoDS against the same published rubric. The deterministic rubric result is our canonical score. The LLM's result is shown here as a sanity check — never mixed into the scoring formula.
| Dimension | Rubric | LLM | Δ (LLM − Rubric) |
|---|---|---|---|
| Pricing transparency | 84 | 20 | -64 |
| Business transparency | 40 | 25 | -15 |
| Shipping clarity | 92 | 85 | -7 |
| Public reviews | 82 | 0 | -82 |
| Product range | 95 | 70 | -25 |
| Access & onboarding | 65 | 40 | -25 |
| Support track record | 78 | 65 | -13 |
| Store integrations | 95 | 75 | -20 |
| Overall | 77 | 45 | -32 |
What this means: Large disagreement — investigate. The LLM read the published signals very differently from the deterministic rules.
Median per-dimension |Δ| = 22.5.
Pricing transparency is low (20) because pricing is visible only after signup/trial. Business transparency is low (25) because the company is private. Review score is 0 because there are no reviews. Product range is 70 because there are 100K+ SKUs. Access is 40 because a free plan is not available. Support is 65 because mostly positive feedback is reported. Integration is 75 because native integrations are available across 4-5 platforms.This is the LLM's own explanation, not editorial commentary from SupplierSpy. The LLM result is a sanity check on the rubric — never mixed into the scoring formula.