Thyris Merchant Services as an Enterprise WhatsApp AI Chat Platform
Thyris Merchant Services is an enterprise WhatsApp AI chat platform that connects store-scoped assistants to Meta WhatsApp Business Messaging. It receives customer messages through a unique webhook, preserves the store and chatbot boundary, and connects the assistant to approved catalog, commerce, and order capabilities.
When to use this scenario
Use this model for customer-facing WhatsApp experiences such as:
- Product discovery and product questions
- Guided selection using store catalog data
- Cart preparation with an explicit confirmation step
- Order information when an approved order tool is available
- Frequently asked questions grounded in merchant content
- Escalation or fallback to a merchant-defined support path
Platform composition
| Capability | Merchant Services role |
|---|---|
| Chatbot instance | Stores the assistant identity, model choice, instructions, and store association. |
| WhatsApp connection | Stores the Meta business account, phone number, status, and channel configuration. |
| Unique callback URL | Receives Meta webhook verification, messages, and status events for one connection. |
| Verify token | Proves ownership during Meta webhook configuration and supports controlled rotation. |
| Catalog and orders | Provide the merchant data used by approved assistant journeys. |
| MCP, ARP, and Router | Expose and restrict agent-accessible commerce tools. |
| Activity signals | Support webhook, conversation, message, and delivery-status troubleshooting. |
Message flow
Required Meta information
Prepare the following values before creating the connection:
- WhatsApp Business Account ID
- Phone Number ID
- Display phone number
- Access token
- App secret
Merchant Services generates a connection-specific callback URL and verify token. Add them to the correct Meta application and subscribe to the messages webhook field.
Implementation journey
1. Prepare the assistant scope
Define the store, supported customer intents, languages, tools, prohibited actions, escalation behavior, and response ownership. Keep the first scope narrow enough to test thoroughly.
2. Create the chatbot
Configure the chatbot identity, instructions, model, and channel. If it uses agentic commerce, connect only the store tools required for the intended journey.
3. Connect Meta WhatsApp
Enter the Meta connection values, then configure the generated callback URL and verify token in Meta App Dashboard. Subscribe to message events and confirm webhook verification.
4. Test channel behavior
Test dashboard events and real inbound messages. An unpublished Meta application may accept dashboard test events while real customer delivery remains unavailable.
5. Validate business journeys
Exercise product found, product not found, ambiguous request, stale inventory, tool denial, downstream failure, unsafe prompt, long response, repeated webhook, and support escalation cases.
6. Launch and operate
Publish the Meta application after its production requirements are complete. Monitor connection status, last webhook time, response failures, delivery status, tool errors, customer fallbacks, and verify-token changes.
Security and privacy controls
- Encrypt Meta access tokens and app secrets at rest.
- Never include channel secrets in prompts or customer-visible messages.
- Verify webhook authenticity using the deployment's configured Meta validation controls.
- Deduplicate retried inbound events.
- Limit tools and data to the selected store.
- Apply customer-data retention and deletion requirements to conversation records.
- Enable mutual TLS in Meta only after ingress client-certificate validation is configured and tested.
- Provide a fast channel-disable path for incidents.
Success measures
- Valid webhook delivery rate
- Assistant response success and latency
- Product discovery or journey completion rate
- Tool denial and tool failure rate
- Fallback and escalation rate
- Undelivered or failed outbound message rate
- Customer requests resolved without exposing unsupported actions