> ## Documentation Index
> Fetch the complete documentation index at: https://docs.cyborg.co/llms.txt
> Use this file to discover all available pages before exploring further.
# About CyborgDB
**CyborgDB** is a Confidential Vector Database: a self-hosted, standalone vector search engine that keeps your embeddings encrypted end to end. It persists to local disk or S3, runs entirely inside your own infrastructure, and adds confidential vector search to your stack **while keeping your data encrypted throughout the entire process**.
### Why *Confidential*?
According to KPMG, 63% of enterprises say that confidentiality and data privacy are their top risk to AI adoption ([1](https://iapp.org/resources/article/consumer-perspectives-of-privacy-and-ai/)). In regulated sectors, this figure increases to 76%. Yet, when it comes to solutions which address these concerns, the market has yet to answer. This leaves a critical mass of companies unserved and unable to adopt AI in their workflows. They need **Confidential AI**.
### Where CyborgDB Fits in Your Stack
Think of CyborgDB as a **dedicated vector layer** that sits alongside your application:
**Self-hosted, with no external dependencies.** CyborgDB persists to local disk or S3 and runs wholly within your own environment. This means:
* **No data leaves your infrastructure** - Run it in your VPC or on-prem
* **Familiar development patterns** - A vector API that slots into existing AI pipelines
* **Minimal operational surface** - Local disk by default; no separate database to run
* **Gradual adoption possible** - Start with vector search, expand as needed
### The Vector Search Problem
Vector Search powers the most popular AI applications - RAG, RecSys, Semantic Search, etc. However, traditional vector databases have a fundamental security flaw: they use (and often store) vector embeddings in **plaintext** to enable fast similarity search.
This creates a massive attack surface. Your sensitive data - converted to embeddings that still contain semantic meaning - sits unprotected in your database. For enterprise AI applications, this is unacceptable.
**CyborgDB solves this** by enabling Approximate Nearest-Neighbor (ANN) search directly over **encrypted vectors**. Your embeddings never exist in plaintext, even during search operations.
### How CyborgDB Works
CyborgDB uses **cryptographic hashing** and **symmetric encryption** to enable vector search over encrypted data:
**The Process:**
1. **Your app sends vectors** → CyborgDB encrypts them using your private key
2. **Encrypted vectors stored** → Your database stores only encrypted data, never plaintext
3. **Search queries encrypted** → Query vectors encrypted with the same key
4. **Search over encrypted space** → ANN search happens on encrypted vectors
5. **Results returned** → Matching vectors and metadata, with vectors remaining encrypted
**Key Innovation:** Traditional vector databases decrypt data during search. CyborgDB performs similarity search directly on encrypted vectors, so your sensitive embeddings are **never exposed in plaintext**.
This approach gives you a **Confidential Vector Database** you run yourself, with embeddings that are never exposed in plaintext.
### Why Developers Choose CyborgDB
**🔄 Works with Your Existing Stack**\
No need to rewrite applications. CyborgDB exposes a familiar vector API that drops into the AI pipelines and frameworks you already use.
**🔒 True End-to-End Encryption**\
Vector embeddings stay encrypted throughout their entire lifecycle - at rest, in transit, and **during search operations**. No plaintext exposure, ever.
**⚡ Production-Ready Performance**\
GPU-accelerated search with CUDA support. Optimized algorithms deliver fast results without compromising security.
**🛠️ Developer-Friendly APIs**\
Familiar programming interfaces and extensive framework integrations (LangChain, etc.) make adoption seamless within existing AI workflows.
**📈 Flexible Deployment**\
Deploy as embedded libraries for maximum control, or as a service for easier scaling. Choose what works best for your team and infrastructure.
### Next Steps
*Get hands-on in minutes*
Choose your deployment path and build your first confidential vector search application
*Embedded vs Service*
Understand the trade-offs between embedded libraries and service deployment
*Memory, Disk, S3*
Learn where CyborgDB persists your encrypted index, and how to pick per environment