AI Features

Day One’s AI features are designed to enhance your journaling experience by processing the content of individual or selected journal entries. These tools use AI to analyze the text you’ve written and provide tailored suggestions, like Go Deeper prompts, to help you reflect and gain deeper insights. 

For the current Labs features, only the specific entry or selected journal entries are processed by AI, and  the AI query  is securely transmitted to the third-party AI services we use without being stored on our servers or theirs. This involves sending the decrypted content of your entry through HTTPS (an encrypted transport protocol) to a service running an AI model, which analyzes the text and provides suggestions. The AI query is deleted after processing, ensuring your entry remains private. The AI model services used may be hosted by our own company, Automattic, or may be a third-party service like OpenAI. In any case, no user content will be used to train AI models except in individual cases where explicit permission has been granted by the user.

No. Day One’s AI features do not access your journal unless you choose to enable that access. By default, AI only processes the content you actively choose to use with an AI feature. If you don’t use AI, none of your journal data is accessed by AI. As we introduce additional AI capabilities in the future, you will remain in control of what AI can or cannot access. Any expanded access will be optional and permission-based.

We understand that using AI in journaling might feel like a big step, especially for something as personal as your journal. It’s natural to have questions or concerns about privacy and security when considering using AI tools like this.

At Day One, we believe AI has the power to transform how people journal for the better. Whether through generating dynamic prompts, summarizing your thoughts, or even transcribing audio recordings, we’re convinced that AI can help make journaling more accessible, intuitive, and rewarding. AI can serve as a creative companion to help you reflect, uncover new insights, and explore your thoughts in more meaningful ways. For those who have used it, Day One’s AI features have unlocked a whole new dimension of journaling. 

Most importantly, your privacy is a top priority. We’ve built this feature with the same care for security that Day One is known for, ensuring your journal entries remain yours and yours alone.

Day One has a deep respect for your privacy and designed our AI features with that in mind. When you use AI features within Day One, the content from an entry or selected entries is temporarily decrypted on your device for processing, and transmitted securely via encrypted protocols (HTTPS) to our servers, and then to an AI service. It is never stored on our servers or those of the AI services we use. Additionally, the AI services we use do not use your content for training purposes.

Please note that this means content from end-to-end encrypted journals is temporarily not end-to-end encrypted while being processed with AI. This is one of the reasons new AI features are opt-in.

In cases where we can utilize an AI model that can be downloaded and run directly on-device, your content will remain on the device during processing, without needing to temporarily bypass end-to-end encryption.

Your content is processed solely to generate content within the specific entry you’re working on, and we never store or share your AI queries without your consent. We’ve built this feature with the same care and security that Day One is known for, ensuring your journal entries remain yours and yours alone. 

We aim to make AI-generated prompts and suggestions as relevant and helpful as possible, but AI language models are a new technology undergoing rapid development and improvement. We’re in the early days, so the results you receive might not always feel aligned with what you’re looking for. Your feedback is invaluable and helps us improve the feature in future updates.

AI features within Day One are opt-in (off by default), and can be disabled anytime via Settings > Labs > AI features. Once disabled, the feature will no longer process any of your entries, and all AI-related interactions will be hidden from Day One.

Day One’s AI features work best if you’ve written a few sentences or more into an entry, as it uses the context of what you’ve written in an individual entry to tailor its suggestions, like Go Deeper prompts. If you’re not sure where to begin, don’t worry—we provide a few starting prompts to help you get going.

Right now, Day One’s AI features are available as part of Day One Labs (a Premium feature,) which allows you to test experimental features. Future availability and pricing will be determined based on user feedback and further development.

At Day One, your journal entries are yours and yours alone. Protecting your content is at the core of everything we build, which is why end-to-end encryption is one of our most important privacy features. But how does it work with AI? Let’s break it down. (Heads up, this is about to get geeky.) 

End-to-End Encryption vs Standard Journals:
One of Day One’s most valuable privacy features is the ability to store your journal entries with end-to-end encryption. This is our default method for all new journals, provided the user has set up or entered an encryption key.

For users on Apple devices, Day One automatically generates an encryption key for new users and securely stores it in iCloud—out of reach of our servers. This ensures that your entries are encrypted both in storage and during transfer, making them inaccessible to us.

If an encryption key isn’t available, new journals are created without end-to-end encryption. However, these journals are still protected with secure transfer protocols like HTTPS and robust security practices on our servers.

AI and End-to-end Encryption:
End-to-end encryption ensures that your journal entries remain completely private. Our servers can store and sync encrypted content across devices but cannot access or read it. This means server-side AI models also cannot access your content directly.

For AI features to work with encrypted content, content must first be temporarily decrypted on your device before being sent securely to the third-party AI service for processing. Once the AI service provides its output, decrypted content is deleted from our servers and form the servers of AI providers. 

Local AI VS Server AI:
AI models are powerful tools but require significant storage space and computing power to run effectively.

Here’s how they differ:

Local AI: Local AI models are compact enough to run directly on your device. This allows your content to be processed privately, never leaving your device—even without an internet connection. Although current local AI models are limited in their capabilities, we’re excited about their potential. 

Server AI: Server-side AI models are much larger and more powerful, running on dedicated computers in data centers. To use server AI, content must be sent securely to the servers for processing. This approach provides greater AI capabilities but requires temporary transmission of content. Think of it like the early days of computing with shared time on powerful mainframes.

Our AI features are currently powered by server AI, but we’re exploring local AI models to maximize privacy. As  AI models become more efficient, we aim to transition more AI-powered capabilities from servers directly to your device, keeping your content as private as possible.

We’d love to hear your thoughts and suggestions on Day One’s AI features. Your feedback is essential in shaping the future of these tools and ensuring they meet your needs. Share your input here.