Built on Horizen Base L3

Confidential AI inference,
guarded by Harpocrates.

Send encrypted prompts, enclave runs inference, get verifiable outputs. No one, not even us, can see your data.

Privacy-first AI infrastructure | Metered in ETH on Horizen L3

Encrypted Input
TEE + ZK Enclave
Verified Output

Why privacy-first AI?

Traditional AI infrastructure exposes your most sensitive data to unnecessary risk

Your AI provider sees everything

Every prompt, every document, every trade secret flows through their servers unencrypted. Complete data exposure to third-party providers.

Regulatory and compliance risk

PII, PHI, financial data, and confidential business information require strict privacy guarantees that traditional AI APIs cannot provide.

No verifiable execution

You simply trust their API black box. No proof of correct computation, no guarantees about data handling or model integrity.

How Harpocrates works

Three steps to confidential, verifiable AI inference

01

Encrypt

Encrypt your prompt or embeddings client-side before transmission. Your data never leaves your control in plaintext.

02

Compute in Silence

Trusted execution environments (TEE) and ZK attestations guarantee privacy and correctness inside a secure enclave.

03

Verify & Settle

Verify cryptographic proofs on-chain and settle compute costs in ZEN on Horizen Base. Fully auditable and transparent.

Privacy-first infrastructure

Everything you need to build confidential AI applications

End-to-end encrypted prompts

Client-side encryption ensures your prompts and data remain confidential throughout the entire inference pipeline.

TEE + ZK attestation

Hardware-backed trusted execution environments combined with zero-knowledge proofs for cryptographic verification.

Horizen Base L3 integration

Native integration with Horizen's privacy-first appchain for secure settlement and proof verification.

ETH-based metering and billing

Transparent, deterministic billing in ETH on Horizen L3. Pay only for what you use with on-chain settlement.

Developer-friendly SDKs

Simple JavaScript and Python SDKs with local encryption helpers and copy-paste examples for common workloads.

Compliance-ready private inference

Meet GDPR, HIPAA, and enterprise compliance requirements with verifiable private computation.

Built for developers

Integrate confidential AI inference into your applications with just a few lines of code.

Simple JS and Python SDKs
Deterministic billing in ETH on Horizen L3
Local encryption helpers
Copy-paste examples for common workloads
example.ts
// Initialize Harpocrates client
const client = new Harpocrates({
apiKey: "your-api-key",
})
// Perform confidential inference
const result = await client.infer({
model: "llm-secure-7b",
input: encrypt("Analyze this...")
})
Powered by Horizen

Built on Horizen Base

Harpocrates leverages Horizen's privacy-first L3 appchain infrastructure for secure settlement, proof verification, and ZEN-based metering.

Horizen Base L3

Privacy-first appchain optimized for confidential computing

ZEN Settlement

Native ZEN token for gas, metering, and billing

Privacy Infrastructure

Confidential computing and zero-knowledge proofs

Use cases

Confidential AI for your most sensitive workloads

Private document analysis

Keep board decks, contracts, and internal memos in sacred silence. Analyze sensitive documents without exposing content to third parties.

Confidential embeddings search

Build semantic search over proprietary data while maintaining complete privacy. No leakage of business intelligence.

Enterprise AI copilots

Deploy AI assistants that work with sensitive customer data, financial records, and confidential communications securely.

Secure ML for regulated industries

Healthcare, finance, and legal sectors can leverage AI while meeting strict compliance and privacy requirements.

Token & Economics

Simple, transparent economics powered by ETH on Horizen L3

Pay per inference in ETH on Horizen L3

Simple, transparent pricing. Pay only for the compute you use with deterministic on-chain settlement.

Throughput tiers with funding controls

Fund with ETH to unlock higher rate limits and preferential pricing as your usage scales.

Future: AI Compute Marketplace

Model providers can offer their models and earn ETH on Horizen L3 in a verified private marketplace.

Roadmap

Our path to production-ready confidential AI infrastructure

1
Phase 1In Progress

Testnet MVP

Initial testnet deployment with encrypted inference capabilities and basic SDK support for early developers.

2
Phase 2Q2 2026

Mainnet Launch

Production mainnet deployment with ETH billing on Horizen L3 (migrating to ZEN when live), plus enterprise-ready features.

3
Phase 3Q4 2026

AI Compute Marketplace

Decentralized marketplace for third-party model providers, staking mechanisms, and advanced governance.

Frequently asked questions

Everything you need to know about Harpocrates