Skip to main content

​
How It Works

​
Web Sources

When enabled, the system automatically processes your query to generate search-optimized terms, queries web sources through our search integration, and returns the most relevant and credible results that support the AI’s response.

​
Smart Suggestions

Our suggestion engine analyzes the conversation context and AI response to generate natural follow-up questions that encourage deeper exploration, clarify complex topics, or explore related areas of interest.

​
Basic Usage

​
Enable Sources and Suggestions

const response = await fetch('https://connectapi.answerr.ai/functions/v1/chat', {
  method: 'POST',
  headers: {
    'Authorization': `Bearer ${YOUR_API_KEY}`,
    'Content-Type': 'application/json'
  },
  body: JSON.stringify({
    model: "gpt-4o",
    messages: [
      {
        role: "user",
        content: "What are the latest developments in quantum computing?"
      }
    ],
    includeSources: true,
    includeSuggestions: true
  })
});

​
Enhanced Response Format

{
  "id": "chatcmpl-abc123",
  "model": "gpt-4o",
  "choices": [
    {
      "message": {
        "role": "assistant",
        "content": "Recent quantum computing developments include significant breakthroughs in error correction, new qubit technologies, and commercial applications..."
      }
    }
  ],
  "sources": [
    {
      "title": "IBM Achieves Quantum Error Correction Milestone",
      "url": "https://research.ibm.com/blog/quantum-error-correction-2024",
      "snippet": "IBM researchers demonstrate a breakthrough in quantum error correction that could enable practical quantum computers...",
      "domain": "research.ibm.com",
      "relevance_score": 0.94
    },
    {
      "title": "Google's Quantum AI Makes Major Advance",
      "url": "https://ai.googleblog.com/quantum-breakthrough",
      "snippet": "Google's quantum team reports significant progress in quantum supremacy experiments...",
      "domain": "ai.googleblog.com", 
      "relevance_score": 0.89
    }
  ],
  "suggestions": [
    "How do quantum error correction techniques actually work?",
    "What commercial applications are closest to reality?",
    "How do IBM and Google's approaches to quantum computing differ?",
    "What are the main challenges still facing quantum computers?"
  ]
}

​
Sources Feature

​
Intelligent Search Processing

The system transforms user queries into effective search terms by:
  • Entity extraction: Identifying key topics and concepts
  • Query optimization: Reformulating questions for better search results
  • Context awareness: Using conversation history to refine searches
  • Relevance filtering: Selecting sources that directly support the AI response

​
Source Quality

Our search integration prioritizes:
  • Authoritative domains: Academic institutions, government agencies, established organizations
  • Recent content: Prioritizing up-to-date information when timeliness matters
  • Relevance scoring: Matching source content to query and response topics
  • Diversity: Including multiple perspectives and source types

​
Source Applications

​
Fact-Checking and Verification

Automatically verify claims in AI responses with supporting web sources, helping users validate information and build confidence in AI-generated content.

​
Research Enhancement

Transform AI responses into research starting points by providing credible sources for further investigation, academic citations, and expert perspectives.

​
Current Events and News

For topics requiring up-to-date information, sources ensure AI responses reflect the latest developments rather than training data cutoffs.

​
Professional Applications

Legal, medical, and technical professionals can verify AI suggestions against authoritative sources before making important decisions.

​
Suggestions Feature

​
Context-Aware Generation

Suggestions analyze multiple factors:
  • Conversation flow: Natural progressions from current topics
  • Complexity levels: Questions appropriate for the user’s apparent expertise
  • Exploration paths: Different angles and related topics worth investigating
  • Clarification needs: Areas where the AI response might benefit from elaboration

​
Suggestion Types

​
Clarifying Questions

“Can you explain that concept in simpler terms?” “What does [technical term] mean exactly?”

​
Deeper Exploration

“What are the implications of this for [related field]?” “How has this changed over the past decade?”

​
Comparative Analysis

“How does this compare to [alternative approach]?” “What are the pros and cons of each method?”

​
Practical Applications

“How would I implement this in practice?” “What tools or resources would I need?”

​
Suggestion Applications

​
Educational Platforms

Help students explore topics systematically by suggesting logical next questions that build understanding progressively and encourage critical thinking.

​
Customer Support

Guide users through troubleshooting processes with intelligent follow-up questions that help identify root causes and explore related issues.

​
Research and Analysis

Provide researchers with systematic exploration paths, helping them consider different angles and ensure comprehensive topic coverage.

​
Content Creation

Help writers and creators explore topics thoroughly by suggesting angles, perspectives, and details they might not have considered.

​
Advanced Features

​
Combined with Compare Mode

{
  "model": "gpt-4o",
  "compareModel": "claude-3-5-sonnet",
  "messages": [...],
  "includeSources": true,
  "includeSuggestions": true
}
Get sources and suggestions for both model responses, allowing comparison of how different models approach fact-checking and conversation flow.

​
Auto-Mode Optimization

Auto-Mode considers whether sources and suggestions are enabled when selecting models, potentially choosing models that work better with external context and follow-up generation.

​
Implementation Example

class EnhancedChat {
  constructor(apiKey) {
    this.apiKey = apiKey;
    this.conversationHistory = [];
  }

  async chat(message, options = {}) {
    // Add user message to history
    this.conversationHistory.push({
      role: 'user',
      content: message
    });

    const response = await fetch('https://connectapi.answerr.ai/functions/v1/chat', {
      method: 'POST',
      headers: {
        'Authorization': `Bearer ${this.apiKey}`,
        'Content-Type': 'application/json'
      },
      body: JSON.stringify({
        model: options.model || 'auto',
        messages: this.conversationHistory,
        includeSources: options.includeSources !== false, // Default true
        includeSuggestions: options.includeSuggestions !== false,
        compareModel: options.compareModel
      })
    });

    const result = await response.json();
    
    // Add assistant response to history
    this.conversationHistory.push({
      role: 'assistant',
      content: result.choices[0].message.content
    });

    return {
      response: result.choices[0].message.content,
      sources: result.sources || [],
      suggestions: result.suggestions || [],
      metadata: {
        model: result.model,
        hasComparison: !!result.comparison
      }
    };
  }

  // Helper method to ask suggested questions
  async askSuggestion(suggestionIndex, lastResponse) {
    if (!lastResponse.suggestions || !lastResponse.suggestions[suggestionIndex]) {
      throw new Error('Invalid suggestion index');
    }

    const suggestion = lastResponse.suggestions[suggestionIndex];
    return this.chat(suggestion);
  }
}

// Usage
const chat = new EnhancedChat(process.env.ANSWERR_API_KEY);

const response = await chat.chat("Explain machine learning algorithms");
console.log("Response:", response.response);
console.log("Sources:", response.sources);
console.log("Suggestions:", response.suggestions);

// Follow up with a suggested question
const followUp = await chat.askSuggestion(0, response);