⚡ Iceberg Database

High-performance serverless vector database built for AI agents and real-time RAG.

🚀 Quick Install

pip install icebergdb

⚡ 30-Second Quickstart

import iceberg

# 1. Connect to Iceberg (reads ICEBERG_API_KEY from environment by default)
db = iceberg.connect(api_key="your_api_key")

# 2. Create an isolated vector collection
db.create_collection("ai_agent_memory", description="Real-time context memory")

# 3. Index text documents or chunks (supports English, Hindi, and multilingual text)
db.index_text("ai_agent_memory", "Iceberg delivers sub-5ms vector queries with zero cold starts.")
db.index_text("ai_agent_memory", "Autonomous AI agents require persistent, low-latency recall memory.")

# 4. Perform hybrid semantic search
results = db.search("ai_agent_memory", query="low-latency agent recall", top_k=3)

for match in results:
    print(f"[{match.score:.3f}] {match.text}")

📁 Uploading Documents Directly (PDF, TXT, Markdown)

# Upload and auto-chunk entire documents in one line
db.upload("ai_agent_memory", "product_spec.pdf")

🔍 Filtered & Scoped Search

results = db.search(
    "ai_agent_memory",
    query="vector benchmarks",
    top_k=5,
    score_threshold=0.4,
    filters={"source": "product_spec.pdf"}
)

🌐 Cloud & Self-Hosted Endpoints

To connect to your local dev engine or dedicated cluster:

db = iceberg.connect(
    api_key="iceberg_secret_key",
    base_url="http://localhost:8000"  # or https://api.icebergdb.in
)

🛡️ License

MIT License © 2026 Iceberg Data Technologies.