Learning Timeline
Key Insights
Response Speed Tips (Latency)
For voice agents, speed is key. Using the 'Gemini 2.0 Flash' model is highly recommended as it is optimized for speed (low latency), making conversations feel more natural without long delays.
The Importance of Data in the System Prompt
Your agent only knows the information you provide. Make sure to include as many business details as possible in the 'System Prompt' so the agent doesn't 'hallucinate' or provide incorrect information to customers.
Prompts
Example System Prompt for Support Agents
Target:
ElevenLabs System Prompt
You are a helpful customer service assistant. We are open from 3 p.m. to 10 p.m. 7 days a week. We are located at 5445 Main Street in downtown Chicago. Answer questions politely and concisely based on this information.
Step by Step
Setting up an ElevenLabs AI Voice Agent (Conversational AI)
- Log in to the ElevenLabs dashboard and navigate to the 'Voice Agents' section.
- Click the '+' (Create Agent) button to start configuring a new agent.
- Select the 'Support Agent' template from the provided list (or choose 'Blank' to build from scratch).
- Set the primary language for your agent using the 'Language' dropdown menu.
- Click on the 'System Prompt' section to input the agent's behavioral instructions and personality.
- Select a language model in the Large Language Model (LLM) section. Choose 'Gemini 2.0 Flash' for fast, high-quality responses.
- Enter specific business details (such as operating hours, address, and FAQs) into the 'System Prompt' text box.
- Click the microphone icon or the 'Test' button to start a voice conversation and test the agent's response.
- Ask trial questions related to the business info you provided to ensure the agent answers accurately.
