AI Transparency
Last updated: April 2026
Tempered (the “Platform”) is a decision-enrichment service provided by Tempered. The Platform uses AI vendors to produce enrichment data — multi-model analysis, dimensional scoring, preflight checks, and outstanding questions — so that the customer, as decision-maker, has more information and guidance before acting. This page explains how AI is used within the Platform, what data flows to AI vendors, and how human oversight is maintained.
How AI Is Used
When you submit an evaluation request, the Platform sends your input to one or more AI vendor APIs for independent analysis. Each vendor assesses the decision and returns structured reasoning, dimensional scores, and outstanding questions. The Platform aggregates these responses into an enrichment verdict that presents the full range of perspectives.
AI is used only for evaluation enrichment — the core function of the Platform. We do not use AI for profiling, automated decision-making about your account, or any purpose beyond what you explicitly request.
What Data Reaches AI Vendors
| Data | Sent to vendors | Notes |
|---|---|---|
| Evaluation request content | Yes | This is the input you submit for enrichment |
| Your identity or account details | No | Requests are anonymised before submission |
| API keys or credentials | No | Authentication is handled server-side |
| Usage history or behavioural data | No | Vendors see only the current request |
AI Vendors
The Platform uses multiple AI vendors to provide multi-perspective enrichment analysis. The specific vendors used for your evaluation depend on the profile selected.
Current vendors include models from Anthropic, OpenAI, Google, and Cohere. Each vendor processes your evaluation input under their respective data processing terms. We select vendors that commit to:
- Not using customer inputs for model training
- Not retaining inputs beyond the processing window
- Providing enterprise data processing agreements
For the full list of sub-processors, see the Sub-processors section of our Data Processing Agreement.
Human Oversight
- All AI outputs are presented as enrichment data, not as directives — the final decision is always yours
- Dimensional scoring highlights where vendors agree and disagree, so you can see the full range of perspectives
- Evaluation results include the reasoning behind each assessment, not just a score
- You can review the full vendor-by-vendor breakdown for every evaluation
Model Limitations
AI models can produce incorrect, incomplete, or biased outputs. The Platform’s mitigations include:
- Multi-vendor consensus: Using multiple independent models reduces the impact of any single model’s blind spots
- Structured output validation: All vendor responses are parsed and validated against a defined schema
- Transparency: Full reasoning chains are shown so you can evaluate the quality of the analysis yourself
The Platform does not guarantee the accuracy of AI-generated enrichment. See our Terms of Service for liability provisions.
Data Retention
Evaluation inputs and AI vendor responses are retained for the duration specified in our Data Processing Agreement. You may request deletion at any time.
Contact
Questions about our use of AI should be directed to privacy@temperedrisk.ai.
- Tempered, a trading style of Your Systems Team Limited (Company No. 06798860)
- ICO Registration: ZC102414
- Registered address: 1 Peach Street, Wokingham, England, RG40 1XJ