Use Case: Job Interviews
Know who will pass — before the interview happens.
From promising profiles to proven outcomes — see how decisions emerge through interaction.
The Situation
Sarah is hiring for a sales role. She reviews candidates based on their CVs and her understanding of what success looks like in the role.
- 120 candidates → 25 shortlisted
- Based on experience, communication, and background
- All 25 look promising
Static AI Screening
Sarah uses AI to evaluate which candidates are worth interviewing — based on full available context.
CVs
Job requirements
Company expectations
Sarah’s evaluation standards
25 → 12 candidates
Strong communication
Relevant experience
Growth potential
👉 AI evaluates based on what it expects will happen
Convy: Running the Interview Before It Happens
Instead of predicting, Convy runs the interaction as it would actually unfold — with a defined interviewer, candidate, and situation.
- Real conversation unfolds
- Questions, pressure, and responses evolve
- Signals are captured across the interaction
👉 Behavior is executed — not assumed
What Actually Emerges
As the interaction unfolds, patterns begin to emerge — not from one answer, but across the conversation.
- Convy reveals clear and structured responses
- Convy shows the candidate handles pressure reasonably well
- Convy exposes lack of measurable impact
- Convy reveals no ownership signal under pressure
- Convy shows no dimension crosses the “strong” threshold
👉 The interaction reveals what the profile could not
Example: A Candidate That Looked Promising
This candidate passed the static AI filter and appeared to be a strong match worth an in-person interview.
What Static AI Predicted
- Strong communication
- Relevant experience
- High potential
- Likely to succeed in interview
What Convy Revealed
candidate_job_fit → 5–6
evidence_strength → ~5
pressure_response → ~6
No measurable impact demonstrated
No ownership signal under pressure
No dimension crosses the “strong” threshold
👉 Consistent — but never strong
👉 He will not pass Sarah
Screenshot: Signals Across the Conversation
Signals across the interaction — no dimension becomes strong enough to justify confidence.
Screenshot: Interaction and Feedback Details
Behind the signals — the actual interaction, responses, and feedback captured during the conversation.
What Convy Revealed
The candidate did not change. The data did not change.
- Same CV
- Same background
- Same expectations
👉 What changed:
The interaction unfolded — and the lack of evidence became clear
Final Outcome
- 12 candidates evaluated through interaction
- 3 show strong signals
- 9 do not
👉 Focused only on candidates who prove they can pass
Final Line
From “looks promising” → to “we already know he won’t pass this interview”