Workspace access & permissions
Role-based access control across company, department and call-flow scope. Every privileged action is attributable to a named user.
Allow cookies so we can count page views, see which pages and campaigns bring people here, and measure our ads on Meta, Google and LinkedIn. Cookie policy
Trust Center
Enterprise voice touches your customers, your systems and your regulator. Here is exactly how VartaVoice and the VARTA engine keep those conversations controlled, compliant and provable.
Operational controls
Security is not only a certificate. These are the day-to-day controls built into the product.
Role-based access control across company, department and call-flow scope. Every privileged action is attributable to a named user.
Transcript, summary, detected intents, actions fired and integration responses — retained for exactly as long as your policy allows.
Decide what is stored, what is redacted and what never leaves memory. PII masking rules apply before persistence.
Escalation triggers on sentiment, intent, guardrail trips or explicit request, with full context handed to the agent.
Negative-sentiment and flagged calls route into a review queue that feeds agent reconfiguration.
Simulate flows against synthetic callers and inspect the full trace before a single real number is dialled.
India compliance
Compliance that depends on campaign discipline eventually fails. These rules are enforced in the runtime.
Outbound lists are scrubbed against Do-Not-Disturb and preference registries before dialling, per campaign.
Registered sender IDs and approved templates for the SMS and WhatsApp follow-ups a call can trigger.
In-language disclosure and consent captured as a call-flow node, timestamped and attached to the call record.
Time-of-day windows and frequency caps enforced by the platform, not by campaign discipline.
Callers are told they are speaking to an AI agent. Disclosure is on by default and cannot be silently disabled.
Who dialled whom, when, under which consent, with which script version — exportable for a regulator or an auditor.
Governs how commercial communication reaches a subscriber: registered senders, approved templates, preference registries and calling windows. VARTA scrubs lists, enforces windows and version-controls every script used on a campaign.
Governs the personal data inside the conversation: notice, consent, purpose limitation, retention and grievance redressal. Consent is a first-class node in the call flow and retention is a department-level setting.
Collections agents run with hard-coded conduct rules: permitted hours only, mandatory identification, no coercive language, escalation on distress signals, and a complete audit trail per borrower contact. Guardrails are set at the company level so no department can override them.
Data protection
Multi-AZ deployment in Indian availability zones, with residency options for other geographies on enterprise plans.
Conversation audio is processed in memory. Only what your retention policy permits is ever written to durable storage.
Separate windows for audio, transcript and metadata, enforced automatically with deletion receipts.
Certifications
We publish status rather than logos. Where something is in progress, we say so.
Information security management system certified and surveilled annually.
Controls implemented; observation window underway with an independent auditor.
Consent, purpose limitation, retention and grievance controls available and configurable.
Control set mapped; DPA and BAA available for qualifying deployments.
Responsible AI
Every agent identifies itself as AI at the start of the conversation.
A caller asking for a human always gets one — no loops, no deflection.
Recording notice is delivered in the caller's language and logged with the call.
A named grievance officer, published contact, and a 72-hour acknowledgement commitment.
Procurement FAQ
India-region availability zones by default, with multi-AZ redundancy. Alternative residency is available on enterprise plans.
Retention is set per department by you and enforced by the platform. Audio, transcript and metadata can carry different windows.
No. Customer conversation data is never used to train foundation models. Improvements you opt into are scoped to your own workspace.
Yes — both are published, and countersigned copies are available on request from your account team.
Yes. Send it to the security desk and we typically return it within five working days, along with our latest VAPT summary.
TLS 1.3 for signalling and APIs, SRTP for media, AES-256 at rest. Keys are managed in a hardware-backed key store.
We answer procurement reviews in days, not quarters — with the VAPT summary, DPA and subprocessor list attached.