The Gemini agent evaluates every resume against the job description in a single API call returning a 100-point match score with full reasoning. One prompt covers scoring, summarisation, and email generation simultaneously.
| Output Field | Type | Description |
|---|---|---|
| Match Score | Number (0–100) | Numeric fit score based on skills, experience, and role alignment |
| Summary | String | 2–3 sentence plain English evaluation of overall candidate fit |
| Strengths | Array | Specific strengths relative to this job description |
| Gaps | Array | Areas where the candidate falls short of stated requirements |
| Recommendation | String | Binary decision — Interview or Reject |
| Email Draft | String | Personalised email — invitation or professional rejection — generated in the same pass |
The original design used two Gemini calls — one for evaluation and one for email generation. This doubled API cost, doubled the failure surface, and added 5 to 8 seconds of latency.
Combining both into one structured prompt eliminated all three problems simultaneously. The email draft quality did not degrade. This is the highest-leverage architectural decision in the entire system.
Resume screening has a higher tolerance for false positives than lead qualification. Missing a great candidate is a worse outcome than spending 30 seconds reviewing one who scored 71 and turned out borderline. The threshold reflects that asymmetry.
Validated through 15+ test resume and JD pairs before locking.