AI Transparency

1. How we use AI

Aimavrik uses AI to help capture and qualify leads, automate workflows, process documents, draft content, and power features of the forthcoming Catalyst platform. AI is a tool we apply where it improves a business outcome — not an end in itself.

2. AI processing lifecycle

At a high level, this is how information moves through an AI-assisted workflow. It is a conceptual overview and does not expose implementation details.

Human review is applied where a decision is meaningful. Logging and monitoring help us keep systems reliable and accountable.

3. Human oversight

AI assists; people decide.

• AI assists by generating drafts, suggestions and automations.
• Humans review meaningful outputs before they are relied upon or shipped.
• Customers remain responsible for business decisions made using AI-assisted outputs.

4. Model limitations

AI models can be wrong, incomplete, biased or out of date. They may produce plausible-sounding but incorrect information. We build in review steps and encourage you to verify important outputs before acting on them.

5. Prompt & input handling

To generate outputs, AI features process the prompts, inputs and context you provide. We limit this to what is needed to deliver the feature and handle it in line with our Privacy Policy. We do not use your content to train AI models.

6. Confidentiality

We treat the data you place in our systems as confidential and apply the security measures on our Security page. Please do not submit sensitive personal data to AI features unless necessary and agreed.

7. Model providers

Aimavrik and the Catalyst platform may use one or more trusted AI providers to deliver services. Providers may change over time to improve quality, availability, security, or performance. We select providers with appropriate security and data-handling commitments, and list the categories we rely on in our Sub-processor List.

8. Customer responsibilities

You are responsible for how you use AI-assisted outputs, for reviewing them, and for ensuring your use complies with applicable law and our Acceptable Use Policy.

9. No professional advice

AI outputs are not legal, financial, tax or medical advice. Consult a qualified professional before making decisions in these areas.

10. Future model updates

AI technology evolves quickly. We may update the models, providers and features we use to improve results, availability or safety, and we keep this page and our Responsible AI and AI Ethics pages current.

11. VoxAudit

VoxAudit is Aimavrik's AI call quality audit platform for BPOs and call centres, operated at voxaudit.in. This section states how AI is used inside it.

• What the AI does. It scores the calls you select against a scorecard you define, per client. It does not decide what good looks like; your scorecard does.
• Outputs include reasons. Every per-parameter score is returned with the reason for it, alongside fatal compliance flags, sentiment, call outcome and coaching feedback.
• A human reviews outputs. Section 3 (Human oversight) applies.
• You can re-audit any call.
• Audio is not retained. Call audio is fetched from your dialer's link at audit time, analysed, and not kept by VoxAudit. Call metadata and audit results are stored. See the VoxAudit annex to our DPA.
• Model provider. VoxAudit uses Google's Gemini API on the paid tier. Google states that content is not used to improve its products. We name the provider here for transparency; the provider is not surfaced inside the product itself. Providers may change over time, in which case this section and our Sub-processor List are updated.