Content Filtering & Safety
Automatic Content Protection
Every query is automatically filtered before processing to ensure quality and safety. Our AI-powered content filter blocks spam, harmful content, and off-topic queries with 98% accuracy.
🛡️ What Gets Filtered
Filter Response Format
When a query is blocked, you'll receive:
{
"error": {
"code": "CONTENT_FILTERED",
"message": "Query blocked by content filter",
"category": "spam",
"confidence": 0.98,
"reason": "Commercial promotional content detected",
"suggestions": [
"Ask a genuine astrology question",
"Remove promotional URLs and content",
"Focus on birth chart or prediction topics"
]
}
}Filter Categories
Valid astrology query - processing continues
Commercial/promotional content - blocked
Dangerous or harmful content - blocked
Non-astrology question - blocked with suggestions
Too vague or ambiguous - needs clarification
Empty, too short, or malformed - immediate block
Performance Impact:
- Basic validation: <1ms (instant)
- AI filtering: under 2 seconds (Vedika Lite 0.9)
- Optimized for speed and accuracy
- Fail-open design: Allows queries on error (UX-first)
Quality Assurance Scoring
Automatic Response Evaluation
Every response is automatically scored across 6 quality dimensions. Responses scoring below 6/10 are automatically enhanced with Vedika Pro 1.0 and re-scored to ensure excellence.
Quality Score Dimensions
Correctness of astrological calculations and interpretations
How well the response addresses the user's question
Coverage of all relevant aspects and details
Ease of understanding for the target audience
Level of insight and advanced analysis provided
Practical remedies and actionable spiritual advice
Enhanced Response Schema
All responses now include comprehensive quality metadata:
{
"answer": "Your personalized astrology response...",
"confidence": 0.85,
"followUpSuggestions": [...],
"qualityScores": {
"overall_score": 8.5,
"accuracy_score": 9,
"relevance_score": 9,
"completeness_score": 8,
"clarity_score": 9,
"depth_score": 8,
"spiritual_guidance_score": 9,
"confidence": 0.92,
"strengths": [
"Accurate planetary details",
"Clear explanations",
"Practical guidance"
],
"weaknesses": [
"Could include more transit timing"
],
"missing_elements": [],
"enhancement_suggestions": [],
"requires_revision": false,
"revision_priority": "low"
},
"metadata": {
"totalStages": 7,
"completedStages": 7,
"processingTime": 52000,
"contentFiltered": true,
"filterCategory": "astrology",
"enhanced": false
}
}Auto-Enhancement Process
- Generate Response - Initial response created by 6-agent swarm
- Quality Scoring - Response evaluated using Vedika Standard 2.1 (under 5 seconds)
- Enhancement Check - If overall_score < 6/10, enhancement triggered
- Enhancement - Vedika Pro 1.0 improves response (5-8 seconds)
- Re-scoring - Enhanced response re-evaluated
- Selection - Best version (original or enhanced) returned
Performance Optimization:
- Quality scoring: Automatic for every response
- Smart enhancement: Only applied when needed (10% of queries)
- Intelligent caching: Up to 40% faster responses for common queries
Advanced Features
LLM Response Caching
Intelligent caching system reduces costs by 40-60% while maintaining response quality. Responses are cached for 30 minutes with SHA-256 key generation.
How Caching Works
- Key Generation - SHA-256 hash from model + prompt + parameters
- Cache Check - Before API call, check if response exists
- Cache Hit - Return cached response (<5ms response time)
- Cache Miss - Call AI model API, cache response for reuse
- TTL Management - Responses expire after 30 minutes
Performance Benefits
Cache Hit Response Time
vs 2-5 seconds API call
Cost Reduction
Based on query patterns
Max Cached Items
~50MB total memory
Cache TTL
Balance freshness & savings
Vedika Intelligence Models
Vedika automatically selects the optimal AI model based on your query complexity and analysis requirements. Our 6-agent swarm architecture provides 82% higher accuracy than generic AI.
Automatic Model Selection
Our intelligent routing system analyzes your query and automatically chooses the best combination of models for speed, accuracy, and cost-efficiency. No configuration needed.
Vedika Pro Ultra
Our most advanced model for complex astrological synthesis, final quality scoring, and deep interpretations
Vedika Swift
Optimized for parallel processing of specialized analyses (birth chart, transits, compatibility, yogas)
Vedika Astro Engine
Swiss Ephemeris powered astronomical calculations with 0.01° precision for all planetary positions
💡 How It Works
- • Intelligent Routing: Query analyzed automatically to select optimal model combination
- • Parallel Processing: Up to 6 agents work simultaneously for 5-8x faster responses
- • Cost Optimization: Swift models handle 70% of work at lower cost without sacrificing quality
- • Quality Control: Pro Ultra model provides final synthesis and verification
- • Smart Caching: 90% cost savings on repeated astrological contexts
💰 Token-Based Pricing
Vedika uses transparent token-based billing. You only pay for what you use, with detailed breakdowns of every query showing exact token usage and costs.
Pricing for 1 Million Tokens
* Actual cost varies based on query complexity, number of agents activated, and token usage
📊 Response Metadata Example
{
"usage": {
"input_tokens": 425,
"output_tokens": 1100,
"cached_tokens": 1575,
"total_tokens": 3110
},
"cost_breakdown": {
"astronomical": {
"calculations_performed": 6,
"cost_usd": 0.17
},
"ai_processing": {
"queen_agent": {
"model": "vedika-pro-ultra",
"role": "lead",
"cost_usd": 0.009
},
"worker_agents": [
{
"model": "vedika-swift",
"role": "specialist",
"cost_usd": 0.003
}
],
"total_cost_usd": 0.02
},
"total_cost_usd": 0.19,
"total_price_usd": 0.52
},
"wallet": {
"balance_before": 20.00,
"deducted": 0.52,
"balance_after": 19.48,
"currency": "USD"
}
}Vedika Astrology Engine
Complete Vedic coverage - Our proprietary astrology engine provides access to 108+ authentic Vedic astrology features powered by Swiss Ephemeris calculations.
FEATURES Comprehensive Features
- Complete birth chart analysis
- 60+ traditional calculations
- Doshas, yogas, divisional charts
- Compatibility & dasha systems
- Multi-agent analysis
- Context-aware routing
- Automatic feature detection
- Parallel computation
Available Calculations
COMING SOON Vedika Open Astrology Engine
We're building an open-source astrology computation engine - like LLaMA for astrology calculations.
What's Coming:
- Free & Open Database: Complete Vedic astrology calculation library
- Self-Hosted Option: Run Vedika engine on your own servers
- Community Driven: Contribute algorithms, yogas, and interpretations
- Enterprise Edition: Advanced features with commercial license
- Multi-Language: Python, JavaScript, Rust implementations
FAQ & Support
Frequently Asked Questions
How do I get an API key?
Sign up at our login page, then generate your API key from the dashboard. For production use, generate a live key (vk_live_*). Test keys (vk_test_*) are deprecated and may be disabled in production.
What's the difference between test and live API keys?
🚨 Test Keys Deprecated: Vedika uses an industry-standard live-keys-only model for production. Test keys (vk_test_*) may return 403 Forbidden when DISABLE_TEST_KEYS=true.
Live Keys (vk_live_*): Required for production use. Token-based billing with pay-as-you-go pricing. Generate from your dashboard.
Test Keys (vk_test_*): Available only in development environments where test keys are explicitly enabled.
What AI models power Vedika?
Vedika uses our proprietary AI models: Vedika Pro 1.0 (enhancement), Vedika Standard 2.1 (primary), Vedika Lite 0.9 (fast queries), and Vedika Express 1.5 (fallback). These are intelligently routed based on query complexity.
How accurate are the astrological calculations?
Vedika integrates Swiss Ephemeris calculations with 108+ Vedic astrology features. Every response undergoes automatic quality scoring across 6 dimensions, with auto-enhancement for scores below 6/10.
What's the typical response time?
Simple queries: 2-5 seconds. Complex analysis: 10-30 seconds. Quality-enhanced responses: 15-45 seconds. The API timeout is 120 seconds (2 minutes).
How does the credit system work?
Each API call deducts credits based on complexity: Chart generation (3 credits), Vedika query (1 credit), Compatibility (2 credits). Credits are automatically deducted and tracked in real-time.
What happens if I run out of credits?
API calls will return an error when credits reach zero. You can top up credits through your dashboard (payment integration coming soon).
Can I use this in my mobile app?
Yes! Vedika API works with any platform that can make HTTPS requests: iOS, Android, React Native, Flutter, web apps, etc.
What error codes should I handle?
Common errors: 401 (Invalid API key), 429 (Rate limit), 400 (Invalid request), 500 (Server error). All errors include descriptive messages.
Is there a rate limit?
Yes. Test keys: 10 requests/minute. Live keys: 100 requests/minute. Enterprise plans available for higher limits.
Do you have SDKs or client libraries?
Currently, we provide REST API endpoints. SDKs for Node.js, Python, and PHP are coming soon. You can use any HTTP client library.
How do I report bugs or request features?
Use our live chat support (bottom-right), email us at support@vedika.ai, or create a support ticket from your dashboard.
Need More Help?
Can't find what you're looking for? Our support team is here to help!
Live Chat
Click the chat icon (bottom-right)
Email Support
support@vedika.ai
Documentation
Browse all sections →