Name: Cencori
Rating: 5 (1 reviews)

RAG (Retrieval-Augmented Generation) combines memory search with AI generation for context-aware responses.

How It Works

User sends a query
Relevant memories are retrieved
Context is injected into the prompt
AI generates a grounded response

Basic Usage

Codetext

const response = await cencori.ai.rag({
  model: 'gpt-4o',
  messages: [{ role: 'user', content: 'What is the refund policy?' }],
  namespace: 'docs'
});
 
console.log(response.content);
// "Based on your documentation, the refund policy allows returns within 30 days..."
 
console.log(response.sources);
// [{ id: '...', content: '...', score: 0.95 }]

Parameters

Parameter	Type	Required	Description
`model`	string	Yes	Chat model
`messages`	array	Yes	Conversation
`namespace`	string	Yes	Memory namespace
`limit`	number	No	Max context items (default: 5)
`threshold`	number	No	Min relevance score
`filter`	object	No	Metadata filter

Custom System Prompt

Codetext

const response = await cencori.ai.rag({
  model: 'gpt-4o',
  messages: [{ role: 'user', content: 'What is the refund policy?' }],
  namespace: 'docs',
  systemPrompt: `You are a helpful customer support agent. 
    Answer questions based on the provided context. 
    If you don't know, say so.`
});

With Metadata Filter

Codetext

const response = await cencori.ai.rag({
  model: 'gpt-4o',
  messages: [{ role: 'user', content: 'How do I get a refund?' }],
  namespace: 'docs',
  filter: {
    category: 'refunds',
    language: 'en'
  }
});

Manual RAG

For more control, implement RAG manually:

Codetext

// 1. Search for relevant context
const context = await cencori.memory.search({
  namespace: 'docs',
  query: 'refund policy',
  limit: 5
});
 
// 2. Build context string
const contextText = context
  .map(r => r.content)
  .join('\n\n');
 
// 3. Generate with context
const response = await cencori.ai.chat({
  model: 'gpt-4o',
  messages: [
    {
      role: 'system',
      content: `Answer based on this context:\n\n${contextText}`
    },
    { role: 'user', content: 'What is the refund policy?' }
  ]
});

Best Practices

Practice	Description
Chunk content	Split long documents into smaller pieces
Include metadata	Add source, date, category for filtering
Set threshold	Filter low-relevance results
Use citations	Show users where answers came from