How Should Users Integrate Transcend AI Agents with Transcend’s products?

Transcend AI Agents integrate with TranscendConnect through three main layers of APIs: (1) external REST/SMPP/HTTP-style APIs on gateways, (2) Kafka topics for events and commands, and (3) internal Java container APIs (SIP Servlets and JAIN‑SLEE) when you want deep, in-platform logic.

1. High-level integration patterns

For most real deployments, the best pattern is:

• Bot as a standalone service
  • Runs as a web service that exposes REST/webhook endpoints (for channels like web, WhatsApp, SMS, voice).
  • Talks to TranscendConnect modules over REST/HTTP, SMPP (for SMS/A2P), and Kafka for events/commands.
• TranscendConnect as the telecom edge and event backbone
  • Border Gateway terminates SIP/SS7/Diameter and pushes structured events into Kafka; it also exposes management/config APIs.
  • A2P Gateway and SMSC expose SMPP and/or HTTP REST APIs for sending/receiving messages and delivery reports.

2. APIs for Border Gateway + Transcend AI Agents

Typical integration for network‑aware Transcend AI Agents:

• Kafka integration
  • Border Gateway publishes call/signaling/events (CDRs, DLRs, alerts) to Kafka topics; the bot service consumes these topics to get real-time context and produces “action” messages (e.g., “notify user,” “throttle route,” “open ticket”).
  • Use native Kafka client libraries from the bot, or a REST Proxy if you want HTTP-only integration into Kafka.
• REST / management APIs
  • Use REST endpoints on your Border Gateway / OSS wrapper for: creating/modifying routing rules, querying subscriber or partner status, triggering tests (pings, call traces), or pulling KPIs that the bot surfaces conversationally.
  • This keeps the bot stateless and lets Border Gateway remain the system of record for configuration and policy.
• Internal Java APIs (advanced option)
  • If you want the “bot brain” to partly live inside the platform:
    • Expose services using SIP Servlets for SIP signaling flows (Java SIP Servlet API).
    • Use JAIN‑SLEE SBBs and Resource Adaptors to consume SS7/SIP/Diameter and drive logic that calls out to external AI/bot services over HTTP.
  • This is more complex but gives ultra‑low latency and carrier‑grade behavior.

3. APIs for A2P Gateway + Transcend AI Agents

For messaging‑centric Transcend AI Agents (campaigns, customer service, alerts):

• SMPP API
  • Use Transcend SMPP endpoints to send/receive SMS; the Transcend AI Agent connects as an SMPP client (ESME) for high‑throughput A2P traffic.
  • Delivery receipts and MO messages are received via SMPP PDUs; you map these to Transcend’s AI Agent internal events.
• HTTP/REST messaging API
  • Use the HTTP/JSON or HTTP/XML SMS/A2P API for simpler integration from bot backends.
  • Inbound messages and delivery reports are pushed as webhooks to the bot, which updates conversation state.
• Campaign / template APIs
  • If your A2P gateway exposes REST endpoints to create campaigns, templates, sender IDs, etc., let the bot call those so users can configure campaigns in conversation (“Create a campaign to send X to segment Y”).

4. Recommended default architecture

For most new work we recommend:

• Let the Transcend AI Agent live outside the telecom stack as a microservice.
• Integrate via:
  • HTTP REST (management/config, low‑volume messaging),
  • SMPP (high‑volume A2P/SMS), and
  • Kafka (real-time events + control plane for Border Gateway and other modules).