Settings

Settings are split into five tabs: Profiles, General, Models, Style, and Appearance. Use Profiles for birth data and saved profiles, General for language/sync/account preferences, Models for interpretation model choices, Style for how the assistant answers, and Appearance for the interface theme.

Most toggles and selectors apply and save immediately. The SAVE PREFERENCES button at the bottom saves the page-level text fields: goals, excluded topics, and current location. Profile fields themselves (name, birth data) are saved by the SAVE PROFILE button on the Profiles tab.

Profiles tab

All saved profiles of your account live here — your own plus family, friends, or clients.

Element	What it does
Active profile	The profile that charts, transits, and assistant answers are computed for
Name	Name of the selected profile
Group	A label for organizing profiles (family, friends, clients...)
Birth date / time / place	Birth data of the selected profile
Rename profile ⟷ Save as new	Switch: what happens when you save with a changed name
SAVE PROFILE	Write the changes
DELETE	Delete the selected profile

How it works:

The active profile is the one personal pages (/profile, /transits, /hd, /returns) and chat use by default. Switch it — and everything computes for that person.
Change the name and save — the profile gets renamed (switch set to "Rename profile").
"Save as new" creates a separate copy instead; the original stays untouched.
Your own profile cannot be deleted — it holds your account's birth data; you can only edit it.
Groups are just organization: moving a profile between groups never changes its birth data. You can manage groups from chat too.

Current location

The Profiles tab also holds your current location and timezone — where you are now (not your birth place!). They affect moonrise, the lunar day, transits, Human Design, and answers about the current sky. Start typing a city — coordinates and timezone resolve automatically.

General tab

Section	What's inside
Language & interface	Site and assistant language; language selector in the header
Astrology	House system (Placidus, Koch...); lunar day by moonrise
Privacy	Save settings to the cloud; dialog analysis; publish artifacts to the public feed; encrypt dialogs
Behavior	Restore last results; auto-refresh on date change
Usage	Daily and monthly request budget, activity graph
Telegram	Bot link status, link to account management
Export / Import	Move settings between devices (JSON), refresh from cloud

Highlights:

Save settings to the cloud governs settings only: when on, your settings (language, models, style, deck, …) are shared across your devices and the Telegram bot; when off, each device and each Telegram chat keeps its own. Profiles, history and results are always saved.
Lunar day by moonrise matters if you follow a specific tradition — schools count the lunar day differently.
Publish artifacts to the feed sets visibility when saving and during some Share actions; it does not publish every new draft. Check beforehand: saving can be public if no preference has been set.

Models tab

Interpretation model choices by purpose. A model only affects the style and depth of interpretation — the calculations themselves are always deterministic and model-independent. The list evolves, so trust the labels in the UI.

The chat model and its effort — and the answer language (any language the site offers), house system, lunar day by moonrise and names in answers — can also be switched right in the conversation: "Set the model to GPT 5.4 Nano", "Set effort to high", "Set the Koch house system" — including several settings in one phrase; see the chat guide.

Block	What it controls
Chat	Main /chat model, temperature (creativity), reasoning level
Insight	The INSIGHT button model
Vision	Image analysis
Images	PNG generation mode, quality, style, format
Remix theme	Visual theme of the AI background (Mystical, Cosmic, Ancient Scroll...)
Tarot deck	Deck and back-side variant

Templates (Chromium, Pillow, SVG) do not charge an AI image generator; text preparation can cost separately. AI modes create a background or the whole picture depending on the surface. Speed is not guaranteed. More: PNG export.

Style tab

Style defines how the assistant answers: tone, address, humor, depth of analysis. It never affects the calculations, only the delivery. It's layered: pick a persona preset and stop there, or fine-tune every field.

Personas

The quick way to set the whole manner of answering at once: choose up to two personas — psychologists (Jung, Rogers...), philosophers (Socrates, Nietzsche, Laozi...), Advaita teachers (Ramana Maharshi, Tolle...), or writers (Borges, Tolkien...). "Synthesis" presets blend a whole tradition. Any field below can be overridden manually — manual settings win over the persona.

A persona can also be picked without opening settings — right in chat: "Set the Jung persona", "Add Tolkien as well", "Remove the personas". Works on the site and in the Telegram bot — see the chat guide.

How to choose a lens, pair two personas and tune the style is covered in detail in the Personas and lenses guide.

Manual controls

Control	What it changes
Tone	Gentle / Direct / Mixed
Formality	How official the answer sounds
Form of address	Informal or formal address; grammatical gender
Spiritualness	Depth of esoteric topics
Motivational	How many suggestions and calls to action
Empathy	Level of emotional support
Humor	From "no jokes" to "unrestrained"
Psychology	Depth of psychological analysis
Philosophy	Depth of philosophical asides
Tradition Focus	Priority system: astrology, Tarot, runes, I-Ching, Human Design...

Goals, topics, follow-ups

Goals (comma-separated) — what you want to achieve ("career, relationships, health").
Opt-out topics (comma-separated) — what not to discuss.
Preference questions in chat — let the assistant ask a short follow-up when your request is ambiguous.
Use names in answers — whether saved-profile names appear in the answer text; off means the assistant phrases answers without names.
Style polish — The extra text-model pass adds time and usage depending on the model and text length.

Example setups

Concise and practical: Tone — Direct, Motivational — Low, Humor — None, no personas.
Warm and supportive: persona Rogers, Empathy — High, Tone — Gentle.
Analytical: persona Jung, Psychology — Deep.
Traditional/contemplative: persona Laozi, Spiritualness — High, Tradition Focus — I Ching.

Appearance tab

Pick the interface theme by clicking a tile — it applies instantly, nothing to save.

Nine themes: Light, Parchment, Arctic, Spring Garden, Rose Quartz (light) and Cosmic Night, Obsidian, Aurora, Midnight Violet (dark).

Common issues

Issue	Fix
Wrong time in results	Check the current timezone on the Profiles tab
Charts computed "for the wrong person"	Check which profile is active
PNG takes long	Models → Images → pick Chromium or Pillow
Answer style is off	Style tab: personas, tone, formality
Goals/topics/location didn't save	Click SAVE PREFERENCES at the bottom of the page
Profile changes didn't save	Click SAVE PROFILE on the Profiles tab

Controlling generation costs

Check each Models block: Chat, Insight, Vision and Images are separate choices. A cheap Chat model does not determine the others' usage.
Choose an image mode explicitly before the first PNG. Selectors save immediately; review the whole block after a change.
Start with concise facts and a template. For AI images, check quality and format. A new preview is not necessarily a free confirmation before generation.
Style polish adds a text-model pass. Preparing long Insight text for PNG can also consume text budget separately, even with template rendering.
Download an existing file when you want the same picture. A fresh export can generate again. Check Usage.

Publication, names in images and image model are different settings. Check all three before saving or sharing. More: PNG, Practical Guide.