Frequently Asked Questions (FAQ)

ZeebeeAI Chat SDK supports all major platforms including web, mobile (iOS and Android), and desktop applications. We provide official SDKs for:

• JavaScript/TypeScript (Node.js and browser)
• Python
• Java/Kotlin
• Swift
• C#/.NET

Additionally, any platform that can make HTTP requests can use our REST API directly.

ZeebeeAI offers several advantages over using AI providers directly:

• Unified API: Access multiple AI models through a single, consistent API
• Model Fallbacks: Automatic failover to alternative models if your primary choice is unavailable
• Enhanced Features: Voice chat, streaming, conversation management, and caching
• Policy Engine: Built-in content moderation and safety features
• Cost Management: Optimized routing to balance performance and cost
• Analytics: Detailed usage statistics and performance metrics

No, you don't need to bring your own API keys for the underlying models. ZeebeeAI manages the connections to all supported AI providers. You only need a ZeebeeAI API key, which you can obtain from your dashboard after signing up.

However, for Enterprise customers who want to use their own API keys for billing or compliance reasons, we do offer a "Bring Your Own Key" option. Contact our sales team for more information.

Rate limits vary by subscription tier:

• Free: 10 requests per minute, 500 per month
• Pro: 60 requests per minute, 5,000 per month
• Enterprise: 300 requests per minute, 25,000+ per month (customizable)

Rate limit headers are included in all API responses to help you track your usage. If you need higher limits, please contact our sales team.

All API requests require authentication using an API key. Include your API key in the request headers:

Authorization: Bearer YOUR_API_KEY

You can generate API keys from your ZeebeeAI dashboard. We recommend creating separate API keys for different applications or environments (development, staging, production).

Yes, ZeebeeAI automatically stores conversation history when you provide a conversation_id in your requests. To continue a conversation across sessions:

1. Include a unique conversation_id in your initial chat request
2. Store this ID on your end
3. Use the same ID in subsequent requests to continue the conversation

Conversations are retained for 30 days by default. Enterprise customers can request longer retention periods.

WebSocket API:

• Bidirectional communication
• Single persistent connection
• Ideal for real-time chat applications
• Supports voice chat capabilities
• Better for mobile applications

HTTP Streaming (Server-Sent Events):

• Unidirectional (server to client only)
• Easier to implement in some environments
• Compatible with more legacy systems
• Uses standard HTTP requests
• Better for web applications

Choose based on your specific requirements and application architecture.

You can install the JavaScript SDK using npm or yarn:

// npm
npm install @zeebee-ai/sdk

// yarn
yarn add @zeebee-ai/sdk

Then import and use it in your code:

import { ZeebeeAI } from '@zeebee-ai/sdk';

const zeebee = new ZeebeeAI('your-api-key');

async function chat() {
  const response = await zeebee.chat({
    messages: [
      { role: 'user', content: 'Hello, how can you help me?' }
    ],
    model: 'gpt-4-turbo'
  });
  
  console.log(response.message);
}

Requirements vary by SDK:

• JavaScript/TypeScript: Node.js 14+ or a modern browser
• Python: Python 3.7+
• Java/Kotlin: Java 8+ or Kotlin 1.4+
• Swift: Swift 5.3+, iOS 13+
• C#/.NET: .NET Standard 2.0+, .NET Core 3.1+, or .NET 5+

All SDKs are designed to have minimal dependencies to keep your application lightweight.

Yes, our JavaScript SDK supports both Node.js and browser environments. However, for browser usage, you'll need to handle CORS and API key security properly:

• CORS: Our API supports CORS for legitimate web applications
• API Key Security: Never expose your API key directly in frontend code. Instead, proxy requests through your backend

For production applications, we strongly recommend creating a backend proxy that handles API authentication securely.

ZeebeeAI supports a wide range of models from leading providers:

• OpenAI: GPT-4 Turbo, GPT-3.5 Turbo
• Anthropic: Claude 3 Opus, Claude 3 Sonnet, Claude 3 Haiku
• Google: Gemini Pro, Gemini Ultra
• Cohere: Command, Command-R
• Meta: Llama 2, Llama 3

Available models may vary by subscription tier. Check your dashboard or the /v2/models endpoint for the most up-to-date list.

The best model depends on your specific requirements:

• GPT-4 Turbo: Highest capability, best for complex reasoning and specialized knowledge
• Claude 3 Opus: Excellent for longer contexts and nuanced understanding
• GPT-3.5 Turbo: Good balance of performance and cost for general purposes
• Claude 3 Haiku: Fastest response times, good for simple questions

You can also use our auto model selection, which intelligently routes to the best model based on the query complexity and your cost preferences.

Context windows vary by model:

• GPT-4 Turbo: 128,000 tokens
• Claude 3 Opus: 200,000 tokens
• GPT-3.5 Turbo: 16,000 tokens
• Claude 3 Haiku: 200,000 tokens
• Gemini Pro: 32,000 tokens

Note that larger context windows typically result in higher token usage and costs. We recommend including only the relevant context for optimal performance and cost efficiency.

ZeebeeAI offers subscription plans with monthly API call limits:

• Free: 500 API calls per month
• Pro: 5,000 API calls per month at $29/month
• Enterprise: 25,000+ API calls per month at $99/month

If you exceed your plan's limits, additional API calls are charged at $0.01 per call. Enterprise customers can negotiate custom overage rates.

If you exceed your monthly API call limit, we'll continue processing your requests without interruption, and bill you for the overage at the end of your billing cycle.

You can set up usage alerts in your dashboard to be notified when you reach 50%, 80%, and 100% of your limit. Enterprise customers can also set hard caps to prevent overages if needed.

Yes, we offer volume discounts for customers with high usage requirements. Please contact our sales team to discuss custom pricing for your specific needs.

We also offer special pricing for:

• Educational institutions
• Non-profit organizations
• Early-stage startups
• Open source projects

Yes, you can upgrade or downgrade your subscription plan at any time from your dashboard.

• Upgrades: Take effect immediately with prorated billing for the remainder of your current cycle
• Downgrades: Take effect at the end of your current billing cycle

ZeebeeAI takes data privacy seriously:

• All data is encrypted in transit using TLS 1.3
• All data is encrypted at rest using AES-256
• We do not use your data to train our models
• Message data is stored only for the purpose of providing conversation history
• Enterprise customers can request data retention customization

For more details, please refer to our Privacy Policy.

Yes, ZeebeeAI is GDPR compliant. We serve as a data processor when you use our services, and we have implemented the necessary technical and organizational measures to comply with GDPR requirements.

Key compliance measures include:

• Data Processing Agreements (DPAs) available for enterprise customers
• Data Subject Access Request (DSAR) support
• Right to erasure compliance
• EU-based hosting options (for Enterprise plans)
• Regular security assessments and audits

Yes, ZeebeeAI offers HIPAA compliance for Enterprise customers. This includes:

• Signed Business Associate Agreement (BAA)
• Enhanced security measures
• Access controls and audit logging
• Encrypted storage with strict access policies
• Regular security audits

Please contact our sales team to discuss your HIPAA compliance requirements.

If you're experiencing authentication errors, check these common issues:

1. Verify your API key is correct and active in your dashboard
2. Ensure you're including the Bearer prefix in the Authorization header
3. Check that your API key has the necessary permissions
4. Confirm you're using the correct API endpoint for your region
5. If using a proxy or firewall, ensure it's not blocking or modifying your requests

Example of correct authentication header:

Authorization: Bearer sk_live_abc123def456...

Incomplete responses can occur for several reasons:

1. Token limits: You may be hitting the model's maximum generation length. Try increasing the max_tokens parameter.
2. Connection issues: For streaming responses, network interruptions can cause incomplete messages. Implement retry logic with backoff.
3. Content filtering: The model's content filter may be stopping generation. Rephrase your prompt to avoid potentially problematic content.
4. Timeouts: Very long generations might exceed service timeouts. Break complex requests into smaller chunks.

For streaming APIs, always check for the final message with is_final: true or finish_reason: "stop" to ensure you've received the complete response.

To optimize latency:

1. Use streaming: Implement streaming to show partial responses as they're generated
2. Choose faster models: Models like Claude 3 Haiku or GPT-3.5 Turbo have lower latency
3. Minimize context: Send only the necessary conversation history
4. Use regional endpoints: Enterprise customers can access region-specific endpoints
5. Implement client-side caching: Cache frequent requests to avoid redundant API calls
6. Use connection pooling: Maintain persistent connections when making multiple requests

For latency-sensitive applications, we recommend using WebSockets over HTTP requests.