Frequently asked questions about activation and use of Prism, the matching agent
Quality of suggestions
Will the matches be different from those currently suggested?
It all depends on what the employee asks for. For example, if they request a position that aligns well with their current skills, then the matches suggested will be close to those from the skills-based Matching context.
If the employee asks a question that does not align with an existing Matching context, then the matches suggested will be completely different from those in the suggestions tab.
How can we ensure that the matches will be better?
The conversation and the agent’s preliminary questions will help personalize and refine the employee’s expectations. Analyzing their profile before suggesting matches will also help reduce obvious inconsistencies.
Each suggested Opportunity will be accompanied by an explanation from the agent to provide context, and the analysis of gaps will help the employee understand their positioning and how to act to reduce those gaps.
Upstream, 365Talents tested the quality of its agent’s Matching using data from 5 different clients.
After launch, 365Talents will monitor the evolution of the following KPIs:
- Ratio of positive feedback vs negative feedback
- Decrease in negative feedback on Matching quality via NPS
Is there a risk that the agent creates unrealistic expectations?
The agent is designed to analyze and take skill gaps into account. When the gap is too large, it explains it and indicates that training may be required, or redirects toward a more accessible Opportunity for the employee.
Can the agent encourage unwanted mobility?
The agent suggests mobility based on open positions and assignments available on the platform. It does not know the specific mobility paths of each client and cannot adapt to the specifics of their mobility policy.
Data security
Will conversations be accessible to third parties (Manager, HR, etc.)?
No, in compliance with ISO 42001.
Are my organization’s and my employees’ data secure?
Yes, in compliance with ISO 42001.
We do not train models on employee data, and conversations are not stored by OpenAI.