We Tested Two AI Systems
on the Same 16 Candidates
Here's what happened.
How it works
Three steps to hiring intelligence
Upload your candidate data
Drop in a CSV with candidate details: scores, experience, education, role.
Our ML model predicts instantly
Trained on your company's own hiring history, it returns a decision in under 100ms.
Confidence scores, factors and next steps
See exactly why each decision was made and what to do next.
Performance
The speed gap is not subtle
Time to result
Our model is 259× faster. Same accuracy.
Our ML Model
Cost per run
Runs entirely on your infrastructure. No API calls, no tokens.
OpenAI GPT
Cost per run
36,679 tokens per batch · $0.00559 actual. That's $6 per 1,000 runs.
Capabilities
Feature by feature
| Feature | Our ML Model | OpenAI GPT |
|---|---|---|
| Prediction speed | 73ms | 19 seconds |
| Cost per run | Free | $0.006 |
| Confidence score | Yes (e.g. 99.8%) | No |
| Feature importance | Yes | No |
| Works offline | Yes | No |
| Consistent results | Always | Varies |
| Plain-English explanation | Yes (AI-powered) | Yes |
| Learns your company data | Yes | No |
Full Results
Prediction results: all 16 candidates
Highlighted rows = models disagreed
| # | Role | Exp | Education | Tech | Interview | ML Verdict | Confidence | OpenAI | Match |
|---|---|---|---|---|---|---|---|---|---|
| 1 | SalesJunior · Growth | 1yr | High School | 46 | 82 | Pass | 100% | Pass | |
| 2 | Product ManagerEntry · Product | 0yrs | Bachelor's | 63 | 36 | Pass | 100% | Pass | |
| 3 | MarketingMid · Growth | 3yrs | Master's | 68 | 61 | Pass | 100% | Pass | |
| 4 | SalesJunior · Growth | 2yrs | Bachelor's | 74 | 73 | Pass | 100% | Pass | |
| 5 | EngineerEntry · Engineering | 0yrs | High School | 69 | 65 | Pass | 100% | Pass | |
| 6 | DesignerJunior · Product | 2yrs | Bachelor's | 74 | 55 | Pass | 100% | Pass | |
| 7 | EngineerMid · Engineering | 3yrs | Master's | 94 | 52 | Pass | 99% | Pass | |
| 8 | HRJunior · Operations | 1yr | Bachelor's | 60 | 57 | Pass | 100% | Pass | |
| 9 | Product ManagerEntry · Product | 0yrs | Master's | 88 | 85 | Hire | 94% | Hire | |
| 10 | EngineerEntry · Engineering | 0yrs | Master's | 75 | 79 | Pass | 93% | Pass | |
| 11 | MarketingJunior · Growth | 1yr | Master's | 89 | 90 | Hire | 100% | Hire | |
| 12 | Data AnalystMid · Engineering | 4yrs | Bachelor's | 83 | 89 | Hire | 81% | Pass | Disagree |
| 13 | DesignerJunior · Product | 2yrs | PhD | 76 | 62 | Pass | 98% | Pass | |
| 14 | Data AnalystJunior · Engineering | 1yr | Bachelor's | 87 | 58 | Pass | 100% | Pass | |
| 15 | HRSenior · Operations | 10yrs | High School | 85 | 86 | Hire | 100% | Pass | Disagree |
| 16 | Data AnalystSenior · Engineering | 9yrs | Master's | 93 | 71 | Hire | 100% | Hire |
Where they split
The 2 disagreements and why they matter
Both cases reveal the same pattern: our model learned your company's actual hiring behavior. OpenAI follows general industry rules.
Why our model is right: It learned your company hires based on performance, not credentials. A senior candidate with 10 years of experience, a strong interview (86), and a strong tech score (85) matches the profile of successful hires in your history, regardless of formal education. OpenAI defaulted to a general rule that penalizes high school education for senior roles.
Why our model is right: A strong interview score (89) and solid experience (4 years) outweigh the degree level in your company's actual hiring history. Your data consistently shows that interview performance is the #1 signal, and this candidate aced it. OpenAI applied generic weighting that doesn't reflect your culture.
Model insights
What the model learned from your data
Interview Score is #1
The most important hiring signal in your data. A strong interview outweighs almost everything else, including technical scores and credentials.
Technical Score is #2
Close behind interview performance. Candidates with both a high tech score and strong interview were hired at near-100% confidence.
Credentials are #4
Education level matters but ranks below experience and scores. Candidates were rejected with high confidence when both scores were low, regardless of degree.
Full factor ranking
Why together is better
ML Model + OpenAI = complete hiring intelligence
ML Model alone
- Fast and accurate
- Free to run
- Learns your culture
- No plain-English reason
OpenAI alone
- Plain-English output
- 19 seconds per run
- $6 per 1,000 runs
- Generic hiring rules
Combined
- Instant prediction
- Free to run
- Learns your culture
- Plain-English reasons
- Actionable next steps
Open data
Download the raw files
Ready to test it on your data?
Upload your own dataset and see predictions in under 100ms, for free.
Register Now