Configuration

This section covers additional settings that allow you to fine-tune Orvis Voice for your specific use case. For authentication and environment settings, see:

• Token Configuration

• Environment Variables

1. Assistant Behavior Customization

Orvis Voice allows you to modify how the AI interacts with users through various configuration options:

Task-Specific Prompts

The AI's personality and expertise depend on the TASK_SPECIFIC_PROMPT variable. You can define prompts such as:

• Tech Support Assistant – "You are a knowledgeable AI specializing in troubleshooting software issues."

• Blockchain Advisor – "You analyze blockchain transactions and smart contracts to provide insights."

• Mental Health Companion – "You engage in supportive, empathetic conversations and provide mindfulness exercises."

Creativity Level

Adjust TASK_CREATIVE_LEVEL (0-1) to control response flexibility:

• 0.0 – Strict, follows the prompt exactly.

• 0.5 – Balanced, allows some variation.

• 1.0 – Highly creative, generates diverse responses.

---

2. Response Filtering & Moderation

To ensure high-quality responses, Orvis Voice can filter out unwanted or irrelevant content.

Keyword Filtering

You can implement a custom response filter by adding a middleware function that checks responses for specific words or topics.

Example (for developers):

function filterResponse(text) {
  const blockedWords = ["spam", "offensiveWord"];
  return blockedWords.some(word => text.includes(word)) ? "Response blocked." : text;
}

Safe Mode (Coming Soon)

A planned Safe Mode feature will allow automatic moderation of AI responses based on predefined safety parameters.

---

3. Caching & Performance Optimization

Orvis Voice includes caching mechanisms to reduce redundant API calls and improve response speed.

Caching AI Responses

You can enable response caching to store previous AI replies for a set duration:

• Environment Variable: CACHE_EXPIRATION_TIME (default: 600 seconds)

• Reduces API costs and speeds up repeated queries.

Parallel Processing

For high-performance scenarios, you can configure Orvis Voice to handle multiple requests in parallel using worker threads.

---

4. Multi-Agent Mode (Experimental)

Orvis Voice can support multiple AI personalities simultaneously, each with a distinct role.

Example Use Case

• Agent 1: Business Advisor – "Provides insights on market trends."

• Agent 2: Technical AI – "Specialized in blockchain security analysis."

Developers can create multiple agents by defining different prompts and managing them within a session.