Unlimited, Offline, and Private AI Chat with AXIØM.talk

Your completely private A.I. friend

Not everyone needs an Assistant*

Sometimes one just needs a supportive friend to listen, all while maintaining your privacy.

✱ and since Assistants need access to world data as well as your personal data, all in the cloud, they are not very private now are they? 🕵

AXIØM.talk is for AI Conversation

Ax uses a combination of the same on-board LLM that powers Apple Intelligence™ and a custom set of AI Instructions. This allows Ax to chat for as long as as little as needed, with fast responses, and a negligible impact on iPhone® battery life.

Ax is designed for quick conversations whenever you need someone to talk to. Because of that pure focus on conversational safety, and keeping those conversation private, Ax is NOT an assistant. It cannot help with coding, nor can it access any of your device data outside of the Foundational Model and its bespoke instructions.

AXIØM.talk is Private AI and Safe AI, by Design

Ax features multiple levels of guardrails in the training of the Foundational Model itself, the Foundation Model Framework APIs it utilizes, and in the structuring of a set of bespoke Instructions which take precedence over user messages (or prompts).

It can talk about real-world topics, but cannot make unsafe recommendations. You will notice some false-positives, that’s how you know it’s working.

Ax is built for Privacy from the start.

At no point does AXIØM.talk transmit your messages or Ax’ responses to the cloud. Your message history is only uploaded or downloaded by the built-in iCloud or Finder back up systems.

Ax has only the most basic analytics. One event each time you launch the app, and if the model throws a warning to help make improvements. Your messages stay yours.

Supports iPhones 15 Pro and iPhone 16 or later, running iOS 26.1+

AXIØM.talk supports English, French, Spanish, German, Italian, Japanese, Korean, and Chinese (Simplified & Traditional, but is not available in China mainland)

As with all AI products, hallucinations and some unpredictable responses may occur. Please use at your own risk.