Hardware Requirements: Small Scale Deployment (Upto 10 Agents)
This guide outlines the minimum non-production hardware needed to validate core Expertflow CX features. It covers requirements for a non-production trial with fewer than 10 agents.
⚙️ Deployment Prerequisites
See CX Deployment Prerequisites for detailed platform, operating system, and network prerequisites required before deploying Expertflow CX.
Core Components (Mandatory)
The CX-Core platform, which includes the unified routing engine, interaction management, Agent Manager, and Data platform, is mandatory for all deployments.
Component | Purpose | Min. Instances | Recommended Specs (per VM) | Notes |
CX-Core Platform (Combined) | Unified Routing, Agent Manager, ETL, CX Analyzer, external statefulset | 1 VM | 6 vCPU, 12 GB RAM, 150 GB HDD/SSD | Services are co-located (Non-HA) |
Voice and Video Components (If Voice is Tested)
These components are optional if the PoC is purely for digital channels.
Component | Purpose | Min. Instances | Recommended Specs (per VM) | Scaling Notes |
Media Server | Voice Processing, Traditional IVR, Recording | 1 System Service | 4 vCPU, 8 GB RAM, 100 GB Disk | Installed as a system service outside the main Kubernetes cluster. |
Voice AI Gateway | Voice stream forking to ASR/TTS/NLU/LLM | 3 VMs | Required only if testing Conversational IVR/Voicebots. |
Add-on Components (Optional)
Component | Purpose | Min. Instances | Recommended Specs (per VM) | Notes |
WFM (Workforce Management) | Optional Scheduling/Forecasting Backend | 0 (Co-located) | 2 vCPU, 6 GB RAM, 50 GB Disk | Dedicated server for PoC environment. |
Surveys and Campaigns | Outbound Campaigning and Surveys Backend | 0 (Co-located) | 1 vCPU, 4 GB RAM, 50 GB Disk | Deployed on the mandatory CX-Core instance. Resources are additive to CX-Core capacity. |
Third-Party AI/LLM | External ASR/TTS/NLU/LLAMA | 1 VM (Consult Vendor) | Highly dependent on model. May require GPU. | Not managed by the CX Helm chart. Must be provisioned separately. |
💡 Summary for Presales
The PoC environment is typically a single-node deployment for core services. Hardware requirements are driven primarily by the need to separate the resource-intensive voice and bot components.