Appearance

Litebase Storage ​

High-performance storage for AI and data-driven applications.

Overview ​

Litebase Storage is a scalable, immutable key-value storage solution designed for AI-powered applications, large-scale data processing, and multi-modal content management. Whether you're working with real-time AI inference, data retrieval pipelines, or enterprise-grade storage, Litebase provides:

  • ✅ Fast storage & retrieval of structured data, embeddings, and AI-generated outputs.
  • ✅ Multi-modal support for text, images, videos, and metadata.
  • ✅ Low-latency caching for real-time AI and data applications.
  • ✅ Versioning & historical tracking for robust audit trails and reproducibility.

Use Cases ​

1️⃣ AI-Powered Data Retrieval ​

  • Store text embeddings for RAG and AI-powered search.
  • Cache vectorized data representations for fast lookup.
  • Maintain historical AI responses for continuous improvements.

2️⃣ Multi-Modal Data Storage ​

  • Store structured & unstructured AI-generated content.
  • Log inference metadata to track model performance.
  • Enable real-time access to multi-modal AI datasets.

3️⃣ Model State & Experiment Tracking ​

  • Log model parameters, training weights, and fine-tuning steps.
  • Store and query dataset embeddings for AI applications.
  • Maintain version-controlled models for reproducibility.

Usage ​

1️⃣ Install the SDK ​

Litebase Storage supports Deno, Node.js.

sh
# Deno
deno add jsr:@litebaseio/storage

# Node.js
npx jsr add @litebaseio/storage

2️⃣ Store Structured Data & AI Outputs ​

Use Litebase Storage to store AI-generated text, embeddings, and metadata.

typescript
import { open } from "@litebaseio/storage";

const storage = open("storage://ai-data");

const tx = await storage.write([
  {
    key: "data/query_001",
    value: {
      prompt: "Summarize climate change research.",
      model: "gpt-4",
      response: "Climate change is driven by greenhouse gas emissions...",
      timestamp: new Date().toISOString(),
    },
  },
]);

console.log(`Stored data. Transaction ID: ${tx}`);
sh
curl -X POST "https://api.litebase.io/storage/ai-data/write" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{
    "data": [{
      "key": "data/query_001",
      "value": {
        "prompt": "Summarize climate change research.",
        "model": "gpt-4",
        "response": "Climate change is driven by greenhouse gas emissions...",
        "timestamp": "2025-02-20T10:00:00Z"
      }
    }]
  }'

3️⃣ Retrieve Stored Data ​

Quickly fetch stored data for retrieval-augmented generation (RAG), caching, or analytics.

typescript
const response = await storage.read(["data/query_001"]);
console.log("Retrieved data:", response.data);
sh
curl -X POST "https://api.litebase.io/storage/ai-data/read" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{
    "keys": ["data/query_001"]
  }'

4️⃣ Query Historical Versions ​

Retrieve previous versions of stored data for debugging, tracking, or retraining.

typescript
const versions = await storage.versions("data/query_001", { limit: 5 });
console.log("Data version history:", versions.data);
sh
curl -X POST "https://api.litebase.io/storage/ai-data/versions" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{
    "key": "data/query_001",
    "limit": 5
  }'

FAQ ​

Can Litebase Storage handle large AI datasets? ​

Yes, Litebase Storage is optimized for high-performance AI and data applications, supporting large datasets, embeddings, and inference results.

How does Litebase Storage handle images and videos? ​

Litebase supports multi-modal storage, enabling structured metadata tracking alongside AI-generated content for efficient retrieval and auditing.

Can I retrieve previous versions of stored data? ​

Yes! Litebase Storage keeps historical versions, allowing access to previous AI-generated outputs and dataset snapshots.