Getting Started with Graphlit: Your First Project

Welcome to Graphlit! This quickstart gets you from zero to a working AI application in 15 minutes. You'll ingest a PDF, run semantic search, and build a RAG-powered chat interface.

What You'll Build

By the end of this guide, you'll have:

A Graphlit project set up
Content ingested and indexed
Working semantic search
A chat interface that answers questions about your content

Time required: 15 minutes
Prerequisites: Node.js installed, basic TypeScript knowledge

Step 1: Create Graphlit Account (2 min)

Go to portal.graphlit.dev
Sign up (free tier available)
Create an Organization
Create a Project

Save these IDs:

Organization ID
Environment ID
Project ID

Step 2: Install SDK (1 min)

npm install graphlit-client
# or
yarn add graphlit-client

Step 3: Initialize Client (1 min)

import { Graphlit } from 'graphlit-client';

const graphlit = new Graphlit(
  'your-organization-id',
  'your-environment-id',
  'your-project-id'
);

console.log('✓ Graphlit client initialized');

Tip: Store IDs in environment variables:

const graphlit = new Graphlit(
  process.env.GRAPHLIT_ORGANIZATION_ID!,
  process.env.GRAPHLIT_ENVIRONMENT_ID!,
  process.env.GRAPHLIT_PROJECT_ID!
);

Step 4: Ingest Your First Document (3 min)

Let's ingest a PDF about AI:

async function ingestDocument() {
  // Ingest PDF from URL
  const content = await graphlit.ingestUri(
    'https://arxiv.org/pdf/2301.00001.pdf',
    'AI Research Paper'
  );

  console.log('Ingested content ID:', content.ingestUri.id);

  // Wait for processing to complete
  let isDone = false;
  while (!isDone) {
    const status = await graphlit.isContentDone(content.ingestUri.id);
    isDone = status.isContentDone.result;
    
    if (!isDone) {
      console.log('Processing...');
      await new Promise(r => setTimeout(r, 2000));
    }
  }

  console.log('✓ Content indexed and searchable!');
  return content.ingestUri.id;
}

const contentId = await ingestDocument();

What happened:

PDF downloaded
Text extracted using AI
Embeddings generated
Content indexed in vector database

✅ Quick Win: You can ingest any public URL—PDFs, web pages, YouTube videos, even entire GitHub repos. Just pass the URL to ingestUri().

Step 5: Your First Search (2 min)

Now search the content semantically:

async function searchContent(query: string) {
  const results = await graphlit.queryContents({
    search: query,
    limit: 5
  });

  console.log(`\nSearch results for: "${query}"\n`);
  
  results.contents.results.forEach((content, i) => {
    console.log(`${i + 1}. ${content.name}`);
    console.log(`   Relevance: ${(content.relevance * 100).toFixed(1)}%`);
    
    // Show preview from first chunk
    const preview = content.pages?.[0]?.chunks?.[0]?.text?.substring(0, 150);
    console.log(`   Preview: ${preview}...\n`);
  });
}

await searchContent('What is machine learning?');

Example output:

Search results for: "What is machine learning?"

1. AI Research Paper
   Relevance: 94.2%
   Preview: Machine learning is a subset of artificial intelligence that enables systems to learn patterns from data without explicit programming...

Try these queries:

"neural networks"
"deep learning applications"
"training models"

💡 Pro Tip: Graphlit uses hybrid search by default (vector + keyword). You get semantic understanding AND exact phrase matching automatically. Learn more in the Search Guide.

Step 6: Build Your First Chat Interface (5 min)

Create a RAG-powered Q&A system:

async function chatWithContent(question: string) {
  // Create conversation
  const conversation = await graphlit.createConversation('My First Chat');
  const conversationId = conversation.createConversation.id;

  console.log(`\nQuestion: ${question}`);

  // Ask question
  const response = await graphlit.promptConversation(question, conversationId);

  const answer = response.promptConversation?.message?.message;
  const citations = response.promptConversation?.message?.citations;

  console.log(`Answer: ${answer}\n`);
  
  // Show sources
  console.log('Sources:');
  citations?.forEach((citation, i) => {
    console.log(`  [${i + 1}] ${citation.content?.name} (page ${citation.startPage})`);
  });
}

await chatWithContent('Explain the main concepts in this paper');

Example output:

Question: Explain the main concepts in this paper

Answer: This paper discusses three main concepts in machine learning: 1) Supervised learning, where models learn from labeled data, 2) Neural network architectures, particularly deep learning approaches, and 3) Training methodologies including gradient descent and backpropagation.

Sources:
  [1] AI Research Paper (page 3)
  [2] AI Research Paper (page 7)
  [3] AI Research Paper (page 12)

✅ Quick Win: Every answer automatically includes citations—no configuration needed. Citations show which pages/documents were used to generate the answer.

Step 7: Add Streaming for Real-Time Responses (2 min)

Make chat feel like ChatGPT:

async function streamingChat(question: string) {
  console.log(`\nQuestion: ${question}\n`);
  console.log('Answer: ');

  let fullAnswer = '';

  await graphlit.streamAgent(
    question,
    async (event) => {
      switch (event.type) {
        case 'message_update':
          // Print text as it streams
          process.stdout.write(event.message.message);
          fullAnswer += event.message.message;
          
          if (!event.isStreaming) {
            console.log('\n\n✓ Complete');
          }
          break;
          
        case 'error':
          console.error('Error:', event.error.message);
          break;
      }
    }
  );
}

await streamingChat('What are the key findings?');

What you see:

Question: What are the key findings?

Answer: The key findings demonstrate that machine learning models can achieve 95% accuracy when trained on domain-specific data...

(Text appears word-by-word in real-time)

Complete Starter Application

Here's everything combined:

import { Graphlit } from 'graphlit-client';

const graphlit = new Graphlit(
  process.env.GRAPHLIT_ORGANIZATION_ID!,
  process.env.GRAPHLIT_ENVIRONMENT_ID!,
  process.env.GRAPHLIT_PROJECT_ID!
);

async function main() {
  console.log('=== Graphlit Quickstart ===\n');

  // 1. Ingest document
  console.log('Step 1: Ingesting document...');
  const content = await graphlit.ingestUri(
    'https://arxiv.org/pdf/2301.00001.pdf',
    'Research Paper'
  );
  
  // Wait for completion
  let isDone = false;
  while (!isDone) {
    const status = await graphlit.isContentDone(content.ingestUri.id);
    isDone = status.isContentDone.result;
    if (!isDone) await new Promise(r => setTimeout(r, 2000));
  }
  console.log('✓ Document indexed\n');

  // 2. Search
  console.log('Step 2: Searching...');
  const searchResults = await graphlit.queryContents({
    search: 'machine learning',
    limit: 3
  });
  console.log(`✓ Found ${searchResults.contents.results.length} results\n`);

  // 3. Chat
  console.log('Step 3: Asking question...');
  const conversation = await graphlit.createConversation('Demo');
  
  console.log('\nQuestion: What are the main concepts?\n');
  console.log('Answer: ');
  
  await graphlit.streamAgent(
    'What are the main concepts in this paper?',
    async (event) => {
      if (event.type === 'message_update') {
        process.stdout.write(event.message.message);
      }
    },
    conversation.createConversation.id
  );

  console.log('\n\n✓ Done!');
}

main().catch(console.error);

Run it:

node quickstart.js

What You Just Built

Congratulations! You've built a production-grade RAG application with:

✓ Document ingestion - Automatic text extraction and embedding
✓ Semantic search - Find content by meaning, not keywords
✓ RAG chat - LLM answers grounded in your documents
✓ Source citations - Every answer shows which pages it came from
✓ Real-time streaming - ChatGPT-style typing animation

Next Steps

Now that you have the basics, explore advanced features:

Customize Processing:

Workflows Guide - Extract entities, generate summaries
AI Models Guide - Choose GPT-4, Claude, or Gemini

Auto-Sync Content:

Data Connectors Guide - Connect Slack, Gmail, Google Drive, etc.

Advanced Search:

Complete Search Guide - Filters, entity queries, performance

Build Knowledge Graphs:

Knowledge Graphs Guide - Extract people, companies, relationships

Production Ready:

Production Architecture - Multi-tenant, monitoring, scaling

Common First Issues

Issue: "Content still processing after 5 minutes"

Large PDFs (100+ pages) can take 5-10 minutes
Check content state: await graphlit.getContent(contentId)

Issue: "No search results"

Wait for indexing to complete (isContentDone)
Try broader queries first

Issue: "Chat gives generic answers"

Make sure content is indexed
Try more specific questions

Get Help

Documentation: docs.graphlit.dev
Discord: discord.gg/ygFmfjy3Qx
GitHub: github.com/graphlit

Welcome to Graphlit! Happy building! 🚀