Privacy and AI Guardrails

We believe technology should deepen your connection to yourself, not replace it. Odyssey's readings are built on an original Enneagram framework written by Tessa, in her own voice. AI helps make that personal. It doesn't invent the psychology.

A framework at the center

Tessa wrote the source material: how each type uses each strategy, plus wing and subtype notes. When you finish a check-in, AI matches your answers to those pieces and writes them into one plain reading for your week.

Reflection, not authority

Odyssey is here to help you notice your patterns. It isn't the final word on your experience. You are.

Owning our part

We're not claiming to have solved AI's challenges. We're choosing to take responsibility for the part we play in them.

Yours alone

Your check-ins, feeling logs, and reflections are tied to your account. Access rules built into the database mean only you can read or change them. Other members can't see them.

Encrypted

Your data is encrypted while it travels between your device and our servers, and while it's stored.

Never sold

We don't sell, rent, or share your check-ins or reflections with advertisers or data brokers.

Payments stay with Paddle

Card and billing details are handled by Paddle, our payment provider. Odyssey never sees or stores your full card number.

Simple site analytics

We use Google Analytics to count page visits so we know which parts of the app are helpful. It never receives your check-in answers, feelings, or reflections.

Account-linked usage analytics

We also record app usage linked to your account through an internal identifier. The owner can see your signup date and plan, along with setup progress and recorded activity dates. Usage reports include check-in and feeling-log counts, reading-open milestones, and Library article-open counts. They don't display your name or email. They don't include your answers or emotion labels, life contexts or reflections, reading content or article titles. These identifiers are pseudonymous, not anonymous.

Deletion and usage history

When your account is deleted, it no longer appears in the Member usage report. Existing pseudonymous analytics events remain for aggregate reporting under our current deletion rules. This report creates no separate copy of your activity and doesn't extend retention.

Newsletter only if you ask

Your email goes to our newsletter service only when you opt in. You can turn it off in your profile anytime.

Only what the reading needs

When a reading is written, the AI receives your check-in answers, your type details, and the matching framework pieces. Your name and email are never part of it.

Personal details removed

If you add an optional note, names, workplaces, schools, and places are stripped out before it's used. The meaning stays. The identifying details don't.

Not used for training

Odyssey doesn't use your check-ins or reflections to train AI models.

Written once, then saved

Each check-in gets one reading, saved to your account. We don't run AI in the background or keep re-reading your history.

No AI in the Library

Library suggestions come from simple type and topic tags, the same way any website sorts articles. No AI profiles you to choose them.

Diagnose you

No mental health or medical labels, and no claims about trauma. A body sensation is never treated as a symptom.

Make your decisions

It won't tell you to leave a relationship, quit a job, or make any big life choice.

Box you in

Your type describes patterns, not a fixed identity. Readings describe protective and expansive sides without judging either one.

Invent reasons

It can only use what you reported and what the framework supports. If the evidence is thin, it says less.

Make up exercises

The small thing to try at the end of a reading always comes from an approved list of everyday practices.

Step past its limits

If a note mentions self-harm, danger, or a medical emergency, Odyssey stops interpreting it through the Enneagram. You'll see a short message pointing you to real-world support instead.

Using less

We use efficient models and skip unnecessary AI calls. Charts, trends, and the Library run on ordinary code.

Supporting oversight

We support meaningful oversight of AI, including data privacy protections, independent safety reviews, environmental disclosure, and clear accountability for harm.