Use AI like a pro, safely — even if it's your first time.
Next-generation advanced security anyone can use. Even beginners can intuitively build refined, pro-level prompts and use AI like an expert.
You want to use AI, but it feels complicated — and you're worried about your data leaking…
With Protez AI, anyone can use AI intuitively, like a pro.
- Security professionals built in state-of-the-art privacy protection and advanced security, strongly preventing data leaks.
- It also includes a Lab for learning AI and security interactively.
All of this, on just one iPhone.
- Prompt Builder
Engineer-level prompts, made intuitive.
Build professional-level prompts intuitively. Create prompts optimized for each frontier AI vendor's models.
- Wiki
Organize information in a form AI can process, and reuse it.
Organize your own context and domain-specific knowledge systematically in a form AI can understand for easy reuse. Easily bring in existing data with the import feature, and enjoy export and secure backup support.
Screenshots are from a version in development and may differ from the actual app.
How to use AI easily and safely
Download
Get it from the App Store.
Initial setup
Choose beginner mode and no complicated configuration is needed at all. Advanced security settings are also available for enthusiasts.
Try it for free
Download the basic AI package and text generation, translation, simultaneous interpretation, and more run right on your iPhone. It's free.
Privacy & Security
Privacy and security, obsessed over by security professionals.
Multi-layer authentication
- App authentication flexibly combines biometrics (required) with an optional password and security key (FIDO2-compliant, NFC only). Connections to the platform use multi-layer authentication combining mTLS (client certificates), passkeys, and OPAQUE (an authentication protocol that uses cryptography and never transmits credentials directly) — strongly preventing data leaks from unauthorized access.
Multi-layer defense-in-depth architecture with post-quantum (PQC) dual encryption
- Multiple independent defense layers mean a single breach doesn't compromise security. Beyond TLS, communication gets a second, application-layer encryption using post-quantum cryptography (ML-KEM-1024). Mutual authentication via mTLS and dynamic certificate pinning powered by DNSSEC block man-in-the-middle attacks. Data stored in the app on your device is strongly protected by encryption (AES-256-GCM) using a key derived with Argon2id.
Zero-trust architecture
- Data sent to the platform is strictly authenticated, verified, and processed in an environment isolated from the outside. AI inference completes entirely in an environment no one can access, and inference results are encrypted with a public key your device generates automatically until you download them. Third-party services are used only for payments and push notifications — no sensitive information is ever sent to them.
Privacy by design
- No personal information or contact details are required to create an account. Voice data used for speaker identification is analyzed only in your device's temporary memory, and is never sent externally or stored. Inference results on the platform are automatically deleted immediately after download, and the data you send is never used to train AI. You choose the region where your data is processed (data residency), and logs are kept only when a security threat is detected.
Safe AI use
- Guardrail AI detects and blocks harmful content in prompts before it reaches the AI (mandatory on the platform). A human-in-the-loop design lets AI output be checked and corrected at each stage, so mistakes don't compound across repeated inference.
Data this app does not collect or store
- Personal information such as your name or email address at account creation
- Voice data and other biometric information such as biometric authentication data (never leaves your iPhone)
- Data sent for inference and inference results (deleted immediately after inference and after you retrieve them)
- Logs (stored only when a security threat is detected)
About the developer
KOBAYASHI Hiroaki
President & Representative Director (also a full-stack cloud and security engineer)
After accounting, internal-audit, and finance roles at two global Japanese companies, he led new-business launches, international operations, and system development at two small venture firms. He now runs Protez (株式会社プロテジ), a company specializing in cybersecurity.
Key credentials
- CISSPInternational information-security certification ((ISC)²)
- CISAInternational information-systems audit certification (ISACA)
- U.S. CPAU.S. Certified Public Accountant
- MBAKeio (Japan) – HEC Paris (France) double-degree program
FAQ
Is there an Android or desktop version?
Currently, only the iPhone version is released.
What data do you collect?
Your account is managed under a cryptographically random identifier. We don't collect your data, except when a security threat is detected.
Is it paid?
The open-source AI that runs on your iPhone is free to download and use. More advanced AI is available for a fee, securely, on the Protez platform, billed as a single-month contract rather than a subscription, so there are no surprise recurring charges.
Do I need to provide personal information to create an account?
No, none is required.
What AI can I use?
Open-source AI that runs on your iPhone, and open-source AI that runs in the cloud.
Can I use Chat GPT, Claude, and other AI models?
This app lets you build prompts optimized for each AI vendor, but since each frontier AI vendor's subscription plans are usually more economical, we recommend using their subscription plans instead.
Is it okay to use this for work?
For business users, we offer Protez AI Private Edition, where sensitive data used for inference is processed on your self-hosted cloud so we never touch your data. We recommend this edition for business use. For details: https://protez.jp/private
Try using AI simply and safely.
It's never too late to start using AI. With Protez AI, anyone can use it like an expert.