This page describes the current AI-assisted TrustRespond.ai workflow. It is not a regulatory certification.
1. Intended purpose
TrustRespond.ai helps business users prepare answers to security and compliance questionnaires using customer-provided documents and context. It can produce draft answers, source references and confidence indicators. It does not autonomously submit questionnaires, certify compliance or make commitments on the customer's behalf.
2. Model provider
The current hosted workflow uses Google Gemini 2.5 Flash through the Gemini API. Relevant questionnaire context and selected document content may be sent to the provider to generate a suggestion. BYOK is not currently implemented.
3. Human oversight
A user must review the underlying source, correct or reject the suggestion and approve the final response before external use. Missing evidence should remain visible rather than being converted into an unsupported claim.
4. Known limitations
- Generated text can be inaccurate, incomplete, outdated or unsupported by the cited source.
- Confidence indicators are workflow aids, not probability guarantees.
- Tables, scans, ambiguous policies and conflicting documents can reduce extraction quality.
- Customers remain responsible for the accuracy and legal effect of submitted answers.
5. Data handling
Uploaded documents and generated results are currently retained for 90 days unless deleted earlier or agreed otherwise. TrustRespond.ai does not use customer content to train its own public model. Provider processing is governed by the relevant provider and customer terms. See the Privacy Policy and Security Overview.
6. EU AI Act context
The general questionnaire-assistance use case is designed as a human-reviewed productivity workflow. Whether a customer's broader use is subject to additional EU AI Act duties depends on the concrete purpose and deployment context. TrustRespond.ai does not claim the label "EU AI Act Ready", regulatory approval or conformity certification.