Build a Real-Time AI Sales Agent

Learn to build a sophisticated real-time voice sales agent capable of natural conversations with potential customers through this comprehensive workshop. Discover how to create both single-agent and multi-agent systems where specialized AI assistants handle sales, technical support, and pricing inquiries using Cerebras hardware optimizations and the LiveKit Agent SDK. Explore the three-phase conversation process including listening (speech-to-text and end-of-utterance detection), thinking (large language model processing and context pulling), and speaking (text-to-speech with real-time responses). Master the integration of Cerebras, LiveKit, and Cartesia technologies while understanding hardware innovations like memory bandwidth improvements and software innovations such as speculative decoding. Build practical skills in minimizing AI hallucinations through custom context loading, implementing structured business information formats, and creating multi-agent systems with specialized roles for greeting, sales, technical support, and pricing. Gain hands-on experience with LiveKit CLI installation, WebRTC protocols, agent handoff mechanisms, and tool calling capabilities to deploy production-ready voice agents for customer service, sales, and technical support applications.

Syllabus

Introduction to the Workshop & Sales Agent Concept
Workshop Benefits & Starter Code
Introduction to Cerebras and its Hardware
Cerebras Hardware Innovations Memory Bandwidth
Software Innovations Speculative Decoding
Recap of Hardware and Software Optimizations
The Power of Voice Agents in Customer Interaction
Key Capabilities of Voice Agents
Applications of Voice Agents Customer Service, Sales, Tech Support
Inside a Voice Agent: Breaking Down the Conversation
Phase 1: Listening Speech to Text & End of Utterance Detection
Phase 2: Thinking Large Language Model & Context Pulling
Phase 3: Speaking Text to Speech & Real-time Responses
LiveKit Agent SDK for Orchestration
Installing Necessary Packages LiveKit, Cartesia, OpenAI
Cerebras Speed Comparison
Installing LiveKit CLI Optional
Understanding LiveKit and WebRTC
LiveKit Applications Meetings, Call Centers, Websites
Integrating Cerebras, LiveKit, and Cartesia
Step 2: Teaching the AI Sales Agent about the Business
Minimizing Hallucinations with Custom Context
Loading Business Information Pricing, Product Descriptions, Objections
Structured Format for AI Context
Step 3: Creating the Sales Agent Code Walkthrough
Implementing the Sales Agent Class Prompting, Context, Initialization
Defining the on_enter Method for Conversation Start
Step 4: Launching and Running the Agent
Expanding the Sales Agent: Multi-Agent Systems
Types of Specialized Agents Greeting, Sales, Technical, Pricing
Agent Handoff in Multi-Agent Systems
Adding Tool Calling
Conclusion & Q&A