CipherQ gives you an authenticated, encrypted file vault and an on-device AI assistant. Keep your documents secure on hardware you own, discover information with keyword and semantic search, and get verifiable answers with page-level citations—with all processing kept strictly local.
Designed for resource-conscious computing on Arduino UNO Q and Debian Linux.
Store, retrieve, and understand your documents on hardware you control.
Authenticated file storage with verified integrity.
Files are protected at rest with unique individual keys under server master protection. Every upload records cryptographic checksums to verify integrity on download, with full support for uploading, downloading, and deleting files.
Ask questions across your documents with on-device processing.
The on-device assistant runs via an internal service using compact local language models. Document queries and text excerpts are processed locally on your server host for direct, confidential assistance.
Fast keyword indexing across all plans, with semantic vectors on AI Pro.
Locate exact terms with instant full-text search across all plans, or upgrade to AI Pro for local semantic vector search to find passages by concept and meaning on your own device.
Optimized for Arduino UNO Q and Debian Linux systems.
Engineered with streaming memory pipelines, responsive concurrency management, and host disk safety margins to operate reliably on compact edge hardware.
Answers grounded directly in your files with source links.
The assistant answers strictly from retrieved passages in your vault, providing clear citations with document names and page references so you can verify every response.
Inspect every line of code, audit the security model, and self-host freely.
CipherQ is released under the permissive MIT license. Anyone can inspect the code, verify the security boundaries, and deploy on their own hardware.
A transparent on-device pipeline from document upload to evidence-backed answers.
Upload files securely through the authenticated API. Each file is assigned an individual key and encrypted before writing to storage, with integrity checksums recorded automatically.
Authenticated encryption · Per-file keys · Integrity verificationWhen you choose to index a document, text is extracted into page-aware passages within an isolated database. Fast keyword search is available across all plans, while AI Pro adds local semantic vectors.
Full-text search · Local embeddings · User isolationAsk questions in plain language. CipherQ retrieves matching excerpts, supplies them to the local model, and returns an answer citing exact filenames and page numbers.
Local model inference · Page citations · Grounded evidenceDesigned for individuals and teams seeking data ownership, reliable document retrieval, and verifiable answers.
Organize research literature, course notes, and study papers. Search large document collections and query texts with exact page citations to support your work.
Store technical documentation, architecture specs, and development notes in a private vault with fast keyword retrieval and local AI question answering.
Explore deep document archives using semantic search to discover connections across reports and reference materials, supported by direct source citations.
Maintain physical custody of important records, financial documents, and personal notes on hardware you own, with modern search and on-device AI.
The advantages of running intelligence on hardware you control, alongside practical deployment considerations.
Choose the storage quota and AI capabilities that suit your workflow. Every plan includes the same core cryptographic protections.
Perfect for individuals getting started with private file storage and fast search.
Ideal for users who need expanded storage capacity for regular document access.
Designed for users seeking search and on-device AI document question answering.
Upcoming plan offering flexible storage quotas for intensive document workflows.
Note: CipherQ's pricing plans are presented for demonstration purposes. Online payments and self-service subscriptions are planned for a future release. During the hackathon preview, plan access is managed administratively.
| Capability | Free | Premium | AI Pro | Pay as you go |
|---|---|---|---|---|
| Storage quota | 1 GB | 5 GB | 5 GB | Unlimited |
| Encrypted vault | Included | Included | Included | Included |
| Upload, download & deletion | Included | Included | Included | Included |
| Keyword search | Included | Included | Included | Included |
| Semantic search | Not included | Not included | Included | Included |
| Local AI document Q&A | Not included | Not included | Included | Included |
| Daily AI query allowance | — | — | 100 queries / day | Subject to limits |
| Monthly price | ₹0 | ₹100/month | ₹150/month | Coming soon |
CipherQ is engineered with transparent cryptographic boundaries and verifiable local processing. We do not make misleading marketing claims.
CipherQ is not end-to-end encrypted or zero-knowledge. The server holds the master key and decrypts authorized files into server memory during downloads and text extraction. Anyone with host system administrator access can inspect key material.
Files are encrypted with unique per-file Data Encryption Keys using streaming authenticated encryption (AEAD), wrapped under a protected server master key. Full cryptographic specifications are detailed in our technical security disclosures.
User passwords are protected with modern memory-hard hashing. User sessions use cryptographically secure random tokens, and all document and search operations enforce strict tenant isolation.
Search indexing, text embeddings, and language model inference run on the local server host. External telemetry is disabled; document contents and queries are never sent to third-party cloud services.
The database stores basic file metadata (names, sizes, timestamps). When hosting CipherQ outside localhost, operators must configure HTTPS or a secure VPN to protect session cookies and file transfers over the network.
CipherQ protects your data against remote cloud interception, multi-tenant leakage, and third-party AI training. It does not defend against an attacker with root OS access or physical hardware extraction.
Clear answers about CipherQ's architecture and capabilities.
uv, generate your server master key using the key manager CLI, and start the service with Uvicorn. Detailed instructions are available in our documentation.
Start with an encrypted vault on your own hardware, and add local AI document search and question answering when you need it.