PLXYAI Call Center

One studio to design the conversation, run the campaign, and watch every call live — between your Telco SIP trunk and your 3CX agents.

Start Initialzr

Read the documentation

Deployment guide

GitHub

🎛️

Visual Scenario Studio

Seventeen node types on a drag-and-drop canvas — speech, slot filling, LLM turns, branching, sub-flows — with validation findings pinned to the offending node.

📞

Real telephony, stock FreeSWITCH

SIP/RTP against the Cambodia Telco trunk, warm transfer into the 3CX agent pool, recordings and compliance built in — no third-party PBX modules.

📈

Campaigns to insight

Outbound campaigns with contact columns, live monitoring per call, drop-off analysis per node, and conversation analysis reports.

🐳

Deploys as two Docker stacks

Docker is the only requirement — no host JDK, Maven or Node. The telephony half can sit on a DMZ host of its own, with exactly three ports crossing the boundary and every credential a Docker secret.

🔐

Multi-tenant by construction

PostgreSQL row-level security per tenant, a three-axis authorization model, edit locks with named holders, and immutable scenario versions.

🤖

Pluggable AI

ASR, LLM and TTS ride behind a provider SPI — cloud or on-premise — with per-node model options and prompt composition policies.

🛡️

Roles that mean something

Three roles — Admin, Project, Monitoring — evaluated on three axes with named denial reasons, and the live account row decides, not an eight-hour token.

🧠

Ask the docs

A built-in AI assistant runs a local LLM in your browser over WebGPU — grounded in these pages, and nothing you ask ever leaves your device. Open it from the corner of any page.

👋

Built by one pair of hands

Designed, engineered and documented by founder Seng Phirum at PLXY — one system, one voice, a single author's intent.