Skip to main content

Blind Computer

Nillion is building a blind computer - a decentralized network of nodes that stores and processes your and your users' most sensitive data while keeping it encrypted and private from the perspective of application backends and the operators running the underlying infrastructure.

The traditional decrypt-compute-encrypt cycle relies on promises or policies, exposing data during processing. Advanced cryptographic techniques like secure multi-party computation (MPC), homomorphic encryption (HE), and trusted execution environments (TEEs) work directly on encrypted data. This shifts the trust model from "trust us with your data" to "trust the cryptography" — reducing what organizations and infrastructure must be trusted when working with sensitive information.

Start Building​

tip

If you are using an LLM during development (Cursor, ChatGPT, Claude Code, etc.), we strongly recommend providing it with our llm.txt file to help the LLM understand how to build on Nillion. Learn more in our AI-Assisted Workflow guidelines.

Quickstarts​

Start building privacy preserving apps by following one of our quickstarts.

🖥️

Private Compute

Run Docker applications in TEEs with cryptographic attestation using nilCC.

🏛️

Private Storage

Read and write records to an encrypted database using nilDB APIs.

💬

Private LLMs

Run OpenAI-compatible LLMs privately in a TEE within a nilAI node, without exposing user data.

Libraries​

🔐

blindfold

Encrypt/decrypt and secret share data using the blindfold library.

🔍

nilRAG

Provide context to SecretLLM from SecretVault with nilRAG library.

Blind Modules​

A collection of specialized nodes (each running one or more blind modules) provide independent privacy capabilities for the blind computer.

  • nilCC nodes for Private Compute: Nodes perform developer-specified or user-specified workflows within TEEs, enabling general-purpose private computation over sensitive data.
  • nilDB nodes for Private Storage: Data is encrypted and split into secret shares, and these shares are distributed across multiple nodes (typically three) so no single node can view or reveal your original data.
  • nilAI nodes for Private LLMs: Individual nodes run AI models inside of a TEE. You connect to a node of your choice and perform private AI inference without the node viewing your data in unencrypted form.

Developer Solutions​

While you can interact directly with nodes, Nillion provides developer-friendly solutions that handle the cryptographic complexity for you.

  • Private Compute offers a unified API and web dashboard that handles TEE provisioning, attestation generation, and workload lifecycle management - allowing you to deploy any Docker application as a secure computation without modifying your code.
  • Private Storage uses the secretvaults SDK to automatically manage multiple nilDB nodes, encryption, and orchestration - so you don't have to manage and coordinate individual nodes manually.
  • Private LLMs provide OpenAI-compatible APIs that abstract away TEE complexity, letting you add private inference to existing applications with minimal code changes.

These solutions handle multi-node coordination, encryption, and complex cryptographic operations automatically, making privacy-preserving development as simple as traditional app development.

Why Build a Blind Computer?​

The biggest breakthroughs happen when people can collaborate on sensitive data without exposing their secrets, enabling exciting new use cases.

  • Healthcare Research: Pool medical data from hospitals worldwide for clinical research without compromising patient privacy.
  • AI Development: Train better models using aggregated data while keeping individual datasets completely private
  • Government Policy: Use real-time social data to influence policy and help vulnerable populations without breaking privacy.
  • Financial Coordination: Enable coordinated trading strategies without front-running risks.