The agent processes a wide range of image types common in insurance interactions: insurance ID cards for policy lookup, photos of property or vehicle damage for claims initiation, medical bills and Explanation of Benefits documents, repair estimates, police reports, contractor invoices, policy declarations pages, and certificates of insurance. Each image type triggers the appropriate downstream workflow — a damage photo routes to claims intake while an ID card photo triggers a policy detail lookup.

When a policyholder sends a damage photo, the agent initiates a structured First Notice of Loss flow within the same WhatsApp conversation. It asks follow-up questions about the date, time, and location of the incident, whether third parties were involved, and the extent of the damage. The photo and all collected data are packaged together and routed to your claims management system as a complete FNOL submission. This approach captures visual evidence at the moment of loss, which is significantly more reliable than descriptions recalled days later during a phone call.

Yes. Tars connects to Guidewire, Duck Creek, and other insurance-specific platforms through Zapier and webhook integrations, alongside native connections to HubSpot, Salesforce, and Zendesk. Image data and structured claim information captured by the agent map directly to your claims system's intake fields. The integration ensures adjusters receive a complete file — including photos — without manual re-entry or reformatting from WhatsApp into your internal systems.

Tars operates through the official WhatsApp Business API, which provides end-to-end encryption for all messages and media. Beyond WhatsApp's native encryption, Tars is SOC 2 Type 2 certified, ISO certified, HIPAA compliant, and GDPR compliant. All policyholder data — including images containing policy numbers, personal information, and medical details — is encrypted in transit and at rest. Role-based access controls restrict who can view submitted materials, and full audit trails are maintained for state insurance regulatory compliance.

Yes. If an image is too blurry, too dark, or does not contain enough information to process, the agent prompts the policyholder to retake the photo with specific guidance — for example, requesting a closer angle, better lighting, or a photo that captures the full extent of damage. The conversational format makes this a natural back-and-forth rather than a form rejection. If the image still cannot be classified, the agent escalates to a human representative with the image and conversation history attached so the policyholder does not have to start over.

WhatsApp is the dominant messaging platform in Latin America, Europe, the Middle East, South Asia, and Africa — markets where policyholders are far more likely to message than call. In these regions, WhatsApp is often the primary digital communication channel, and deploying insurance service there feels native rather than forced. North American carriers also see adoption, particularly among younger demographics and in markets with significant immigrant populations who already use WhatsApp as their default messaging app. Any carrier serving policyholders who prefer messaging over phone calls benefits from this channel.

The agent scales to handle thousands of simultaneous WhatsApp conversations without degradation in response time or image processing quality. This matters most during catastrophe events — hurricanes, wildfires, widespread hail storms — when carriers face sudden spikes in claims volume. While phone lines experience 45-minute hold times during these events, the WhatsApp agent provides immediate intake capacity for every policyholder who sends a damage photo, ensuring your CAT response team receives clean, image-documented FNOL files from the start.

The AI agent transfers the conversation to a live representative with the complete context attached — the original image, any additional photos, all conversational data collected during the interaction, and the agent's classification of the inquiry type. The human agent picks up exactly where the bot left off, so the policyholder never has to re-explain their situation or resend their photos. This handoff works through integrations with your existing service desk tools, including Zendesk, Salesforce Service Cloud, and HubSpot.