AI Disclosure

How the Eligna AI assistant works

Powered by Claude, running in the EU

The Eligna AI assistant is powered by Anthropic’s Claude models, large language models built by Anthropic. We do not call Anthropic directly. The models are run for us by Google Cloud (Vertex AI) on servers in the European Union, so your message and the context that goes with it stay in the EU. Claude generates a response, which Eligna delivers back to you.

Before your first message

What you tell the assistant can touch on your health or other personal things, so it asks for your explicit consent once, on its first screen, before anything is saved or sent. We record the wording you agreed to and when. You can withdraw it at any time in Account & preferences, which stops the chat and deletes your saved conversations.

What is sent with each message

We build this on our servers from your account, fresh for each message, so the assistant knows what you are working with:

• Your name, and the name of your household if you are in one
• Your statement about who you are, if you have written one, unless you have turned off Remember things about me. A statement that looks as though it mentions your health, beliefs or relationships goes only if you have turned on Remember sensitive details
• Your conversation: the last exchange in full, and short notes on up to 20 exchanges before it, each with what you said, the start of the reply and anything the assistant created
• Your active goals, your habits and whether each one is on a run, and your unfinished tasks
• This week’s reflection, if you wrote one
• What is on your Today page, and, if you have set how much time you have each day, how much of each of the next seven days is already planned
• The names of the people you are due to catch up with or have plans with, from your People area
• Your household’s chores, meals, plants and admin items coming up, if you share a household
• Your money, in totals only (see below)
• How many journal entries you have and when you last wrote, and the summaries of your entries only if you have switched on Journal insights
• How many of your AI credits you have used so far, and when they reset
• Today’s date, and the country your connection comes from, so the assistant can give you the right helpline if you need one

Google keeps the request and the reply for up to 30 days so it can check for abuse of the service, after which they are deleted. Content flagged by an automated safety system, or which the law requires to be kept, may be held for longer.

It answers in your language

The assistant replies in the language you write to it in, even when everything else it is given is in English. When a message has no clear language, such as only a number or a name, it replies in the language you chose in Eligna. Names you gave things, such as tasks, chores, meals and people, stay exactly as you wrote them. Helpline details are copied word for word, never translated.

What is not sent

• Journal entries: never sent to the chat. If you switch on Journal insights in Account & preferences, each new or edited entry is read once, separately, to write a short summary of it, which you can read, edit or delete under the entry. The assistant then uses those summaries, never the full entries. Switching Journal insights off deletes every summary.
• Single payments: never sent. The assistant sees your money in totals only: this month’s spending, how much of each budget is used, how far each savings goal has come, and your net balance in the household. A bill due today can appear on your Today list by its name, never by its amount.
• Other household members’ personal data: Eligna never sends another person’s private habits, goals, tasks, or journal to the AI.
• Where you are, beyond the country: your address and your IP address are never sent to the AI.
• Your password or authentication tokens: these are never sent to the AI.

Your conversations are saved to your account

On every plan, your conversations are saved to your Eligna account, so they follow you to any device you sign in on. We store them in the European Union, and only you can read them: other members of your household cannot. Your device keeps a copy as well, so a conversation opens at once and you can read it offline.

Each conversation is deleted automatically 12 months after its last message. You can delete one sooner from the chat history, or all of them with Clear AI history in Account & preferences, under AI & data, which leaves your tasks, habits, goals, journal and everything else as they are. Deleting your account deletes every conversation.

We also keep how many AI credits you use each month and on what kind of task, to apply your plan’s allowance, and, if you send feedback about a reply, a reference to the conversation it came from.

Free accounts get 20 AI credits a month. Paid plans get 100 a month, for each person. A message to the assistant uses one credit, and so does a photo it reads; journal summaries use one credit for every five entries.

Where your data is processed

Your account data and personal records (habits, goals, tasks, journal entries, household information) are stored in the European Union (Frankfurt, Germany).

When you use Eligna’s AI features, the relevant context for your request is sent to Google Cloud (Vertex AI) for processing in the European Union, on the same continent as the rest of your data. It is not transferred to the United States. Google keeps it for up to 30 days to check for abuse of the service, then deletes it. Neither Google nor Anthropic uses your data to train their AI models.

You can use Eligna without using the AI features. The AI features are optional and only activate when you choose to use them.

Model training

Neither Google nor Anthropic uses this data to train their models. Your messages and the context sent with them are used only to generate a response, and to check for abuse of the service for up to 30 days, and are not used to improve or train any AI model.

For more information, see Google Cloud’s privacy notice (the processor) and Anthropic’s Privacy Policy.

AI responses are not professional advice

The Eligna AI is designed to help with everyday life: habits, goals, tasks, and household coordination. It is not a substitute for medical, psychological, legal, or financial advice.

If you are dealing with a serious mental health issue or personal crisis, please speak to a qualified professional. If you tell the assistant you might hurt yourself, Eligna answers straight away with the crisis and emergency lines for the country your connection comes from, or with lines that work anywhere when it cannot tell.

AI responses may occasionally be inaccurate. Always apply your own judgement, especially for anything important.

Which content is AI-generated

Rather than putting a label on every item, we tell you which kinds of content are generated:

• Illustrations and icons throughout the app
• Meal and plant imagery in the catalogue
• Recipe text is generated with AI assistance and reviewed before publication
• Suggestions and replies from the assistant

Individual images do not carry a label. They depict no real person, place or event: they are illustrations of dishes and plants, so a mark on each one would tell you nothing this page does not.

Plant care notes are the exception in the other direction. They are written by hand, not generated.

Contact

Questions about how Eligna uses AI: privacy@eligna.app