Responsible AI
Last updated: June 29, 2026
AI should make HR teams faster without taking people out of the loop. This statement explains the principles behind how Palette designs, deploys, and governs AI across the platform.
1. Our Principles
Palette builds AI to support people operations, not to replace human judgment. HR decisions affect people's livelihoods, so we hold AI in this domain to a high standard. These principles guide how we design, deploy, and govern AI across the platform.
2. Human Oversight
AI agents assist people; they do not have the final say on decisions that affect individuals. We design workflows so that a person reviews and approves outcomes for consequential actions, and customers can require human approval for sensitive steps. AI outputs are recommendations until a person acts on them.
3. Transparency
We are clear about when you are interacting with AI and what an agent did. Agent actions are logged and attributable, and we explain the basis for recommendations where we can. We do not hide automated processing behind a human label.
4. Data Minimization
Agents access only the data needed for a task, scoped by the requesting user's permissions. We send the minimum necessary context to model providers, and we do not use Customer Data to train public foundation models. Less data in motion means less risk.
5. Security
AI operates inside the same security and permission boundary as the rest of the platform. Prompts and context are protected with the same encryption, access controls, and tenant isolation as other Customer Data. You can read more on our security page.
6. Fairness
We work to reduce the risk that AI produces biased or unfair outcomes in HR contexts. This includes careful feature design, testing, and giving customers controls and human review rather than fully automated decisions. We recognize that fairness requires ongoing attention, not a one-time check.
7. Model Governance
We govern the models and providers we use. This includes evaluating providers, contracting so that Customer Data is not used to train their general models, monitoring performance, and maintaining the ability to change or roll back models. We document significant changes that affect how AI behaves.
8. Customer Control
Customers stay in control of AI in their workspace. You decide which agents are enabled, what they can access, and where human approval is required. You can configure, constrain, or turn off AI features to match your policies and risk tolerance.
9. High-Risk HR Decisions
For decisions with legal or similarly significant effects, including hiring, compensation, discipline, and termination, we expect meaningful human review and do not support relying solely on automated output. We provide guidance and controls to help customers keep a person accountable for these decisions and to meet their own legal obligations.
10. Future Compliance
We track the evolving landscape for AI governance and build toward it. This includes aligning with frameworks such as the NIST AI Risk Management Framework and preparing for obligations under the EU AI Act, particularly for AI used in employment contexts. We will update this statement as our practices and the regulatory environment develop.
11. Contact
Questions about our approach to responsible AI? Contact us at founders@palettetechnology.com or (408) 889-1158.