Skip to main content

Quick Start: Using Mambu Docs with AI Agents

For AI/Agent POCs and Vector Database Ingestion

Generate the Export​

# Clone the repo (if not already)
git clone https://gitlab.com/mambucom/product/knowledge-management/docs/api-docs/api-site-docusaurus.git
cd api-site-docusaurus

# Install dependencies
yarn install

# Export all documentation
yarn export-docs

This creates docs-export.json (~140 MB) with 4,507 documents and 1.6M words.

Upload to S3​

# Compress for smaller upload
gzip -c docs-export.json > docs-export.json.gz

# Upload to S3
aws s3 cp docs-export.json.gz s3://your-bucket/mambu-docs/docs-export.json.gz \
  --content-encoding gzip \
  --content-type application/json

Use with AI Agent​

Python Example (LangChain)​

import json
from langchain.vectorstores import Pinecone
from langchain.embeddings import OpenAIEmbeddings

# Load documentation
with open('docs-export.json') as f:
    docs_data = json.load(f)

# Create embeddings
embeddings = OpenAIEmbeddings()

# Process documents
texts = []
metadatas = []

for doc in docs_data['documents']:
    texts.append(doc['cleanContent'])
    metadatas.append({
        'title': doc['title'],
        'url': doc['url'],
        'tags': ','.join(doc['metadata']['tags']),
        'category': doc['id'].split('/')[0]
    })

# Store in vector DB
vectorstore = Pinecone.from_texts(
    texts=texts,
    embedding=embeddings,
    metadatas=metadatas
)

# Query
query = "How do deposit accounts work in Mambu?"
results = vectorstore.similarity_search(query, k=3)

JavaScript Example​

const fs = require('fs');

// Load documentation
const docsData = JSON.parse(fs.readFileSync('docs-export.json', 'utf8'));

console.log(`Loaded ${docsData.statistics.totalDocuments} documents`);
console.log(`Total content: ${docsData.statistics.totalWords.toLocaleString()} words`);

// Filter for specific topics
const depositDocs = docsData.documents.filter(doc =>
  doc.metadata.tags.includes('deposits')
);

console.log(`Found ${depositDocs.length} documents about deposits`);

// Access document content
depositDocs.forEach(doc => {
  console.log(`\nTitle: ${doc.title}`);
  console.log(`URL: ${doc.url}`);
  console.log(`Content preview: ${doc.cleanContent.substring(0, 200)}...`);
});

JSON Structure​

{
  "meta": { /* Export metadata */ },
  "statistics": {
    "totalDocuments": 4507,
    "totalWords": 1597277,
    "tags": ["api", "deposits", "loans", ...]
  },
  "documents": [
    {
      "id": "docs/deposits/deposit-accounts",
      "title": "Deposit Accounts",
      "url": "https://docs.mambu.com/docs/deposits/deposit-accounts",
      "cleanContent": "Full text content optimized for AI...",
      "metadata": {
        "tags": ["deposits", "accounts"],
        "lastModified": "2026-01-09T12:52:40.704Z"
      }
    }
  ]
}

What's Included​

✅ User Guide (all topics) ✅ API v1 Reference ✅ API v2 Reference ✅ Streaming API ✅ Payments API ✅ All metadata (tags, categories, URLs) ✅ Cleaned content (optimized for AI) ✅ Full markdown (with code examples)

Filter by Topic​

// Find all API v2 docs
const apiV2 = docs.documents.filter(d =>
  d.metadata.tags.includes('api-v2')
);

// Find deposit-related content
const deposits = docs.documents.filter(d =>
  d.cleanContent.toLowerCase().includes('deposit') ||
  d.metadata.tags.includes('deposits')
);

// Get by category
const apiDocs = docs.documentsByCategory['api'];
const userGuide = docs.documentsByCategory['docs'];

Available Categories​

  1. api - All API reference documentation (v1, v2, streaming, payments)
  2. docs - User guide and conceptual documentation
  3. documentation-guidelines - Internal documentation standards
  4. markdown-page - Misc pages

Available Tags (518 unique)​

Common tags: api, deposits, loans, clients, accounts, transactions, configuration, authentication, authorization, webhooks, api-v2, api-v1, etc.

See statistics.tags in the export for the complete list.

Update the Export​

# Regenerate after documentation updates
yarn export-docs

# Or specify custom filename
node scripts/export-docs-for-ai.js my-export-$(date +%Y%m%d).json

Need More Details?​

See the complete Documentation Export Tool guide for:

  • Advanced usage patterns
  • Automation and CI/CD integration
  • Troubleshooting
  • Complete API reference

Questions?​