AI Transparency Statement

Last updated: July 3, 2026

At Klariqo, we believe transparency and accountability are essential when developing and deploying artificial intelligence. This AI Transparency Statement explains how we design, operate, and govern our AI technologies across our product suites.

This document describes our voluntary approach to AI transparency and responsibility. It is not a binding legal agreement. The legal terms governing your use of Klariqo are detailed in our Terms of Service and Privacy Policy.

Our AI Principles

Transparency

We believe that callers have a right to know when they are interacting with an automated system. Every call handled by our Managed Voice AI begins with an explicit automated disclosure. We do not attempt to deceive call participants into believing our conversational systems are human.

Privacy and Security

We protect all data processed by our AI systems. Your data is encrypted in transit and at rest using industry standard cryptographic protocols. Data processed by our AI is strictly partitioned, and we do not permit call recordings or transcripts to be accessed by other clients or exposed publicly.

Reliability

We continuously test our AI systems against a battery of automated guardrails covering conversation safety, response accuracy, and operational guidelines. We monitor system performance in real-time and actively refine our software to address edge cases, latency, and transcription errors.

Fairness

We work to serve all callers equitably, regardless of accent, dialect, or speaking style. The speech recognition and language models we use support a wide range of English accents and speaking styles, and we monitor for and work to reduce disparities in how the system performs across them.

How We Use Data

Managed Voice AI

For clients utilizing our conversational voice agents:

• Real-Time Processing: The AI processes live call audio streams in real-time to transcribe speaker voice, determine intent, generate text responses, and synthesize natural-sounding speech.
• Recording and Transcription: If enabled by the client, the AI generates and stores dual-channel audio recordings and synchronized transcripts of the conversation.
• Data Retention: Managed Voice AI recordings and transcripts are retained for a default period of 90 days, or for a custom duration specified by the client in their dashboard settings.

Compliance & QA scoring AI

For clients utilizing our Compliance and QA Layer:

• Post-Call Processing: The AI ingests completed call recordings from your dialer post-call, transcribes the conversation, separates speaker channels, and analyzes the text.
• Automated QA Scoring: The AI evaluates the transcript against custom scorecards and evaluation rules configured by the client, identifying potential compliance flags and scoring sentiment.
• Signed Record Generation: The AI structures the metadata, transcript, and scores into a standards-based, cryptographically signed virtualized conversation (vCon) record.
• Evidence Storage: These signed vCon records are stored securely in your dashboard environment, where they are retained as active evidence for verification purposes as long as your subscription remains active (see our Privacy Policy for details).

Service Improvement & Opt-Out

• We analyze anonymized call patterns to improve our system's accuracy, response quality, and conversational turn-taking.
• All client data is stripped of personally identifiable information and account identifiers before being used in any technical improvement process. No individual business, caller, or specific call is identifiable within our evaluation data.
• You can opt out. If you do not want your anonymized data used to improve our AI systems, you can opt out at any time while retaining full service functionality. To opt out, contact [email protected].

What We Do Not Do

• We never sell or share caller personal data with third parties for their own marketing or commercial purposes.
• We do not use your proprietary call scripts, configurations, or business data to train models for other clients.
• We do not train machine learning models on identifiable caller information without your explicit consent.

AI Capabilities & Limitations

We believe in being completely honest about the strengths and technical limitations of current artificial intelligence systems.

Managed Voice AI Limitations

• AI models can occasionally misinterpret speech, particularly when callers use sarcasm, slang, or speak over background noise. Our battery of automated guardrails minimizes this, but no conversational system is completely error-free.
• The AI is bound by the parameters, qualification scripts, and business logic configured in your dashboard. It cannot make human judgments outside its defined scope, and it will not provide medical, legal, or financial recommendations.
• Speech recognition accuracy is dependent on telephone line quality and background noise. When the AI is unsure of what was said, it is programmed to politely ask the caller for clarification.

Compliance & QA scoring AI Limitations

• Product Insights, Not Verdicts: Automated QA scores, risk flags, and sentiment analyses represent automated predictive insights. They are designed to streamline quality assurance workflows and highlight areas of concern, but they do not constitute formal compliance certifications or legal determinations.
• No Consent Validation: The AI evaluates what was said during the call. It cannot verify or certify that a consumer's prior express written consent was validly obtained before the call occurred.
• Diarization Confidence: On mono-channel recordings (where both call participants are captured on a single audio track), the accuracy of speaker separation (diarization) may be lower than on dual-channel recordings. The AI will honestly flag lower-confidence speaker attributions within the vCon metadata rather than obscuring them.
• Human-in-the-Loop: We design our Compliance and QA Layer to leverage and assist your internal quality assurance teams. We recommend using automated compliance flags to highlight risk-prone calls for human review, rather than completely replacing human oversight.

Caller Controls

AI Disclosure

Every call handled by our Managed Voice AI begins with a clear disclosure that the caller is speaking with an automated system, helping businesses comply with consumer protection regulations and state laws.

Human Escalation

Callers can request to speak with a human agent at any point during a conversation. The AI is programmed to immediately route the call to your designated human transfer path without argument or delay.

Deletion Requests

• During a live AI call: Callers can say "forget me" or "delete my information." The AI agent will confirm the request and trigger an automated system command to purge stored profile data associated with that specific call session.
• Via email: Callers or consumers can request deletion of their data by contacting [email protected].
• Scope of Deletion: Deletion purges active caller profile data, stored preferences, and call history associated with the caller's phone number within Klariqo's databases.
• Business Records Exception: Deletion requests do not alter or delete records that our business clients are legally required to maintain (such as transactional records, audit logs, or active regulatory compliance evidence). Managing these records remains the responsibility of the client business.

Business Controls

Retention Configuration

Dashboard administrators can configure custom retention windows for call recordings and transcript data to align with their internal privacy policies.

Model Training Opt-Out

You can exclude your anonymized call data from our model refinement processes while retaining full, unrestricted access to all platform features. Contact [email protected] to configure this exclusion.

Call Recording Toggles

For Managed Voice AI, call recording is completely optional and can be toggled on or off in your dashboard settings at your sole discretion.

Version History

Questions & Feedback

We welcome your feedback regarding our AI transparency standards and system operations:

Email: [email protected]