Scope & who we are

This Privacy Policy applies to personal data processed by Codex Labs Corp (“Chatter Genius,” “we,” “us,” or “our”) when you visit the Website (chattergenius.ai) or use our Platform to provision and manage AI Agents across voice, SMS, and email. For Website visitors, we act as a controller. For Platform customers, we generally act as a processor/service provider handling data on our customers’ documented instructions. In specific cases (e.g., billing, account administration, website analytics), we act as an independent controller.

Key definitions

  • Personal data: Information that identifies or relates to an identified or identifiable individual.
  • Customer data: Conversation logs, transcripts, agent configurations, knowledge base documents, and metadata that our customers generate within the Platform.
  • AI Agent: The virtual entity configured by the customer to conduct guided conversations via phone, SMS, or email, optionally grounded in a customer-provided knowledge base.
  • Workflow: A scheduled or on-demand AI Agent conversation using one selected communication channel and one destination.
  • Knowledge base: A collection of customer-uploaded documents that are chunked, embedded, and made available to an AI Agent via semantic vector retrieval.

Information we collect

The data we process depends on your relationship with us:

  • Website usage data: Page views, approximate geolocation, browser attributes, and session engagement metrics collected via Google Analytics 4 (only with your consent). See our Cookie Settings.
  • Contact and newsletter data: First and last name, email address, company, inquiry type, message content, marketing opt-in choice, and the submitting page or source, as applicable. Public form requests also include standard network and browser data such as IP address and User-Agent. Newsletter subscriptions require an email address and include their page or source.
  • Account data: Email address, name, phone number, hashed password (bcrypt), avatar image, mailing address, assigned platform role and access scopes, email and phone verification status, and Stripe customer identifier.
  • Conversation data: Content processed by AI Agents including voice audio streams, SMS text messages, and email bodies, along with associated metadata (timestamps, call SIDs, routing information, message threading identifiers). Voice audio is streamed to the configured realtime provider; the Platform stores final transcript text and related metadata rather than a call recording feature.
  • Agent configuration data: Agent name, role, language, instructions, approved and blocked topics, guardrail configurations, and template selections.
  • Knowledge base data: Documents uploaded by customers (PDF, Word, PowerPoint, Excel, CSV, JSON, HTML, Markdown, plain text) and their derived vector embeddings. Active storage and vector retrieval are scoped by customer and agent.
  • Analytics data: Per-conversation analytics generated by our engine, including sentiment scores, emotion classifications, intent labels, frustration metrics, CSAT estimates, agent quality assessments, compliance flags, language quality scores, topic extractions, and LLM-generated summaries.
  • Billing and usage data: Subscription tier, billing cycle, AI vendor mode (platform-managed or customer-supplied LLM), usage events (call counts and minutes, SMS count, email count, LLM token consumption, and knowledge base storage bytes), monthly usage snapshots, invoices, and Stripe payment metadata.
  • Verification data: One-time passcodes (OTP) sent to your email and phone number during account verification and password reset flows. OTP codes are stored temporarily in Redis and expire after a short period.
  • LLM configuration data: If you elect to use a supported customer-supplied LLM account, we store the API key and endpoint configuration for OpenAI, Azure OpenAI, Google Gemini, Anthropic, or a custom OpenAI-compatible service in encrypted form. Twilio and SendGrid credentials are not currently customer-configurable, and AWS Bedrock credential validation is not currently available.

How we use information

  • Orchestration: To provision AI Agents, allocate communication assets (phone numbers via Twilio, email sender identities via SendGrid), route calls and messages, inject knowledge base context, apply guardrails, and execute scheduled workflows.
  • Security: To authenticate users via RSA-signed JWTs with token rotation and blacklisting, detect fraud, prevent abuse via toxicity moderation, enforce per-IP and per-email rate limits and login lockouts, and apply optional carrier or VoIP screening when configured.
  • Conversation controls: To record non-sensitive guardrail and redaction metadata, redact configured high-confidence PII, credential, and MNPI patterns before text persistence and onward egress, detect common opt-out signals, and apply configured conversation policies. These controls assist with, but do not guarantee, legal or regulatory compliance.
  • Analytics: To process completed conversations through our asynchronous analytics engine and generate available message- and conversation-level signals such as sentiment, emotion, intent, frustration, estimated CSAT, agent quality, compliance, and language quality. Individual enrichments may abstain or return bounded defaults when confidence or a dependency is unavailable.
  • Billing: To aggregate call, voice-minute, SMS, and email usage for subscription allowances and overage calculations, record additional LLM and knowledge-base usage telemetry where available, and manage Stripe checkout, billing portal access, invoices, and webhook processing.
  • Improvement: We use aggregated, de-identified metadata to monitor platform health, optimize latency, and improve service reliability. We do not use Customer Data to train models operated by Chatter Genius. Data sent to third-party AI providers is governed by the selected provider account, configuration, and applicable terms.

AI & LLM processing

Our platform uses Large Language Models (LLMs) to power AI Agent conversations and generate analytics. Here is how data is handled:

  • Provider processing: Platform-managed text, realtime voice, embeddings, moderation, and analytics requests currently use OpenAI services. Retention and model-improvement treatment by a third-party provider depends on the applicable provider account, product, configuration, and terms; this Policy does not promise zero retention by those providers.
  • Guardrails: A local sensitive-value boundary scans user and assistant text before text persistence and onward service egress. Additional deterministic rules, semantic topic checks, moderation, and conversation-safety controls operate according to the active deployment configuration and provider availability. Realtime voice audio reaches the configured provider before the Platform receives the final transcript, so this local boundary cannot retroactively filter audio already processed by that provider.
  • Knowledge base retrieval: When enabled, the platform retrieves semantically relevant chunks from the agent’s knowledge base (stored as vector embeddings in pgvector) and injects them into the LLM context. During ingestion, parsed text chunks are sent to the configured embedding provider for embedding and indexing. At query time, query text is embedded and relevant retrieved chunks are sent to the selected text LLM as response context. Rich document parsing may also use LlamaParse as described below.
  • Document parsing: Plain-text knowledge-base formats are parsed locally. When cloud parsing is configured, rich documents such as PDF, Word, PowerPoint, and Excel files are sent to LlamaParse, a LlamaIndex service, to extract text before chunking and embedding.
  • Analytics processing: Completed conversation transcripts are processed by transformer models (HuggingFace, spaCy, KeyBERT, VADER) for local analytics and by LLM providers for summary generation and agent quality scoring. This processing occurs solely to deliver analytics features to you.
  • Customer-supplied LLM: If you configure a supported OpenAI, Azure OpenAI, Google Gemini, Anthropic, or custom OpenAI-compatible account, relevant conversation data flows to that provider under your account and applicable agreement. Customer-supplied realtime voice is currently limited to OpenAI and Azure OpenAI. Twilio and SendGrid remain platform-managed.

Sharing & disclosure

We do not sell personal data. Current disclosures include:

  • Infrastructure, software, and service providers: We use the following categories to deliver the Website and Platform:
    • Cloud infrastructure: Google Cloud Platform (Cloud Run, Cloud SQL, Secret Manager, Artifact Registry, VPC)
    • Databases: PostgreSQL (with pgvector extension) for relational data and vector embeddings; MongoDB Atlas for conversation and analytics storage; Redis for caching, rate limiting, and session management
    • Communication providers: Twilio (voice calls, SMS); SendGrid (email delivery)
    • AI/LLM providers: OpenAI for platform-managed text, realtime voice, moderation, embeddings, and analytics; and a supported customer-selected OpenAI, Azure OpenAI, Google Gemini, Anthropic, or custom OpenAI-compatible provider when the customer supplies that account
    • Document parsing: LlamaIndex (LlamaParse) for cloud extraction of text from supported rich knowledge-base documents
    • Payment processing: Stripe (subscriptions, invoicing, checkout)
    • Language quality: Self-hosted LanguageTool software running within our Google Cloud environment for grammar, spelling, and style checks in analytics; this is not a separate external LanguageTool recipient
    • Website analytics: Google Analytics 4 (website usage only, consent-gated)
    • Address services: Google Maps Platform and Google Places for address autocomplete and map display; address searches and full saved addresses may be sent to Google when those profile features are used
    • Website asset delivery: Google Fonts and Google Hosted Libraries, plus Amazon CloudFront for the hosted jQuery dependency. These services receive standard browser request data, including IP address and User-Agent, when the Website loads those assets, independently of analytics consent
  • Public avatar storage: In the current Early Access configuration, an avatar you upload is served from a public Google Cloud Storage object URL. Anyone who obtains that URL may access or cache the image. Removing the avatar from your profile detaches its URL from your account but does not immediately delete the stored object.
  • Legal authorities: When required by law, subpoena, or court order, or to protect the safety of our users or the public.
  • Corporate transfers: In the event of a merger, acquisition, or sale of assets, subject to continued privacy protections. We will notify affected users before personal data is transferred to a new entity.

Security

We use technical and organizational safeguards designed to protect the Platform:

  • Authentication: RSA-signed (RS256) JWT access and refresh tokens with automatic rotation, SHA-256 hashing of refresh tokens, Redis-backed token blacklisting, and per-email/per-IP login lockout after failed attempts.
  • Encryption and credentials: External Platform ingress requires TLS 1.2 or later. Platform service credentials are supplied through Google Cloud Secret Manager, customer-supplied LLM keys are encrypted in the application database, and passwords are hashed with bcrypt.
  • Access control: User, administrator, and internal-system roles are combined with owner-or-administrator and scope checks on user-facing resources. The current Platform does not provide customer-managed custom roles or organization membership administration.
  • Network security: X-Frame-Options, HTTP Strict Transport Security (HSTS), trusted host validation, and proxy-aware IP resolution are applied at supported web and API boundaries.
  • Service authentication: Internal API operations use service credentials with optional previous-key windows to support rotation. Twilio, SendGrid, and Stripe callbacks use provider signatures or configured verification tokens, and selected downstream calls use timeouts and circuit breakers.
  • Log hygiene: Application logging filters are designed to redact common secret-like fields and avoid logging raw credentials.

Retention

We retain data only as long as necessary for the purposes described in this Policy:

  • Conversation and analytics data: Retained according to the applicable subscription, order form, Data Processing Agreement, operational requirements, and legal obligations. You may submit a deletion request through the contact method below; requests are evaluated subject to applicable law and required retention.
  • Knowledge base data: The Platform supports deletion of individual active knowledge-base documents and their associated active vector chunks. Residual copies may remain temporarily in backups or provider systems under their applicable retention schedules.
  • Contact and newsletter data: Retained as needed to respond to requests, operate a requested subscription, maintain unsubscribe or suppression status, document consent choices, meet legal obligations, and support reasonable operational needs. Unsubscribing changes the subscriber status but does not automatically delete the underlying record.
  • Account data: Retained for the duration of your account plus any legally required retention period. A removed avatar object may remain at its public object URL and becomes eligible for the current provider lifecycle deletion rule when it reaches 365 days of age; actual provider deletion and downstream cache expiration may occur later.
  • Billing data: Retained for periods required by applicable tax, accounting, payment-processing, and record-keeping obligations.
  • OTP and session data: Temporary data (OTP codes, session tokens, and refresh-token records) expires according to the applicable flow. Verification codes and enrollment challenges are generally valid for 10 minutes, backend access tokens for 30 minutes, and backend refresh-token records for up to 7 days unless revoked or rotated earlier. The signed-in web application uses an encrypted, HTTP-only Auth.js session cookie to hold the backend session credentials; see our Cookie Settings for related browser storage.

Account termination does not necessarily delete every record immediately. We handle deletion, de-identification, backup expiration, and legally required retention in accordance with applicable agreements, provider capabilities, and law.

International data transfers

Our infrastructure is hosted on Google Cloud Platform. Data may be processed in the United States or other jurisdictions where our providers operate. Transfer safeguards depend on the applicable provider and customer agreements and legal requirements. Contact us to request information about mechanisms applicable to your processing, which may include Standard Contractual Clauses (SCCs) or equivalent mechanisms as required by applicable law.

Children’s privacy

The Platform is not directed to individuals under the age of 18. We do not knowingly collect personal data from children. If you believe we have inadvertently collected data from a minor, please contact us at privacy@chattergenius.ai and we will promptly delete the data.

Your privacy rights

Depending on your jurisdiction, you may have the following rights regarding your personal data:

  • Access: Request a copy of the personal data we hold about you.
  • Correction: Request correction of inaccurate or incomplete data.
  • Deletion: Request deletion of your personal data, subject to legal retention requirements.
  • Portability: Request your data in a structured, machine-readable format.
  • Restriction: Request that we limit the processing of your data in certain circumstances.
  • Objection: Object to processing based on legitimate interests.
  • Withdraw consent: Where processing is based on consent (e.g., analytics cookies), you may withdraw consent at any time via our Cookie Settings page or by contacting us.

CCPA/CPRA (California residents): You have the right to know what personal information we collect, request its deletion, and opt out of any “sale” or “sharing” of personal information. We do not sell or share personal information as defined by the CCPA/CPRA.

GDPR (EEA/UK residents): You may exercise any of the rights listed above. Our legal bases for processing include contractual necessity, legitimate interest, and consent. You may lodge a complaint with your local supervisory authority.

To exercise any of these rights, please email privacy@chattergenius.ai. We will respond within 30 days (or sooner as required by applicable law).

Cookies

For detailed information about the cookies and similar technologies we use, including how to manage your preferences, please visit our Cookie Settings page.

Changes to this policy

We may update this Privacy Policy to reflect changes in our practices, technology, or legal requirements. Material changes will be communicated via email or a prominent notice on our Website. We encourage you to review this page periodically.

Contact us

For questions about this Policy, to request a copy of our Data Processing Agreement (DPA), or to inquire about our security practices, please contact: