Getting Started
This guide walks you through the end-to-end process of setting up HubTalk AI, from initial briefing through to production launch.Onboarding Process
Submit a brief
Fill out the onboarding brief describing your use case, requirements, and expected call volumes. The HubTalk AI team will review it and follow up with any clarifying questions.
Demo preparation
A demo is prepared using either your data or publicly available data. Demo preparation typically takes 1–3 business days, and up to one week in some cases.
Pilot agreement
After the demo, you’ll agree on pilot scope: format, topic coverage, and deployment model. For enterprise clients, data security is a key factor — SaaS, private cloud, or on-premise options are available. Each pilot has custom KPIs (e.g., % of dialogs without escalation, SL, FCR).
Platform Overview
Once onboarded, you have access to the HubTalk AI dashboard with the following sections:| Section | Description |
|---|---|
| Agents | Create and manage voice agents — configure LLM, ASR, TTS, prompts, and functions |
| Analytics | View call metrics, conversation breakdowns, and performance data |
| Call Detail | Drill into individual calls with full transcripts, recordings, and export to CSV |
| Knowledge Base | Upload documents (FAQ, product info) that agents use to answer questions |
Setting Up Your First Agent
Navigate to Agents
Open the Agents section in the dashboard. You’ll see all agents for your organization with search by agent name and company.
Create a new agent
Click Create Agent and configure:
- LLM — Select the language model
- ASR — Select the speech recognition engine
- TTS — Select the text-to-speech engine
- Prompt — Write the agent’s system prompt (single prompt mode)
- Greeting — Set a fixed or dynamic greeting message
Add functions
In the right panel, add functions like operator transfer, call ending, or custom API integrations.
Configure post-call analysis
Set up post-call analysis variables to extract structured data from conversations.
Set up webhook
Configure the webhook endpoint to receive call data after each conversation.
Test the agent
Use the built-in chat testing feature to validate your agent’s behavior before going live.
Connect telephony
Work with your telephony team to set up SIP trunks and bind phone numbers to your agent.