Vervoe’s AI helps identify top candidates by assessing their performance in real-world scenarios. The AI learns from your grading and becomes more accurate with every response you grade. Here’s a simple guide on how it works and how you can train it.
How Vervoe's AI Works
Vervoe has a 3 way, machine learning model. Vervoe's preference model mimics human grading, learning from your feedback to assess candidates more accurately.
AI Optimise: Make Sure the Right Candidates Rise to the Top
AI Optimise helps you understand how well your model is performing and when human grading is still needed. It also shows how closely your hiring team’s scores align with the AI. As your team continues to grade, the AI improves, eventually reaching a point where it can handle most of the work automatically.
Why AI Optimise is Useful
- Spot scoring issues early – See when more human input is needed.
- Build trust with your team – Show exactly how the AI is making decisions.
- Reduce manual grading – Once optimised, the AI can handle it for you.
- Ensure fairness – AI applies the same criteria to every candidate.
- Track progress – Watch how your grading improves AI accuracy over time.
Understanding Model Status
As you grade, the AI will move through different learning stages. This helps you know when to stop grading and start shortlisting:
| Status | What it Means | What You Should Do |
|---|---|---|
| Low | Not enough data or graded examples. | Grade more answers and add correct answer samples. |
| Medium | The AI is learning, showing some accuracy. | Keep grading to improve its understanding. |
| High | Performing well with advanced accuracy. | Grade only if lots of new completions come in. |
| Optimised | Highly accurate and stable. | No grading needed unless the role changes. |
What You’ll See in the AI Optimise Modal
When the AI has graded enough responses, a modal will appear. It gives you insights to help guide your next steps:
- Number of completions scored – How many responses the AI has already assessed.
- Model status – See where your model sits on the learning scale.
- Grading recommendations – Whether you should continue grading.
- Hiring Manager Reviews – How closely hiring manager scores match the AI’s.
Training & Optimising the AI Model
Vervoe’s AI identifies top candidates by learning from your grading. With each response you score, the model becomes more accurate and consistent. Here's how to train and optimise it effectively.
✏️ Step-by-Step: Train the AI & Improve Accuracy
| Step | How | Why |
|---|---|---|
| Use a Variety of Assessors | Involve different team members in grading. If you would like to get Hiring Managers involved, share this guide. | Reduces bias by bringing diverse perspectives into the training data, leading to fairer, more balanced scoring. |
| Grade a Range of Responses | Score poor, average, and excellent answers. Don’t just focus on the best. | Teaches the AI to recognise the full spectrum of candidate responses – not just the top tier. Click here to understand the grading scale. |
| Collaborate with your Team | Align with your team on when to stop grading. Use AI Optimise Portal to check how well the model is performing. | Ensures confidence in the AI’s accuracy so you can move from grading to shortlisting. |
Here is a video on how to grade candidate responses.
Once you’ve graded enough responses, the system will display a ‘high’ status on each question – this means the AI has developed a strong understanding of your preferences.
Vervoe's AI continuously learns from each candidate response you grade. If you haven’t reached high status yet, don’t worry—our AI has already learned from every answer you’ve graded so far. The level of AI training you continue to do is entirely up to you.
We recommend collaborating with your internal team to determine the optimal point to stop grading and start shortlisting candidates.
Important: AI scoring is dynamic – scores can change based on new grades, completions, or removed candidates. We recommend waiting 24 hours after any updates before shortlisting.