Settings & Tools

Connection

Start here if you want to connect Ensoul to a model provider. The app supports two backends: OpenRouter and Local (Ollama).

In the Connection card you can switch backend, save your OpenRouter key, or set the Ollama URL. On desktop, the default local address is usually http://localhost:11434. On mobile, the phone must be able to reach the machine running Ollama over the same network.

API keys are stored locally on your device. They are not included in backups, so if you restore a backup later you will need to paste them in again.

Models

The model picker lets you choose which model powers your conversations. With OpenRouter, the picker is grouped by tier so you can compare context window, provider, and cost before you choose.

Model cards show the model name, provider, context size, and cost. The idea is simple: pick a stronger model for deeper conversations, or a cheaper one when you want to keep costs down.

You can also set per-presence model overrides. That means a specific presence can use a different model than the global default. Open the per-presence model sheet, tap a row, and choose an override. Leave it on Global default if you want the shared setting to apply everywhere.

Generation

Generation controls let you shape how the model speaks. These settings change tone and sampling behavior without changing the presence itself.

• Style Preset: choose Auto, Creative, Balanced, or Precise.
• Temperature: higher values make replies looser and more varied.
• Top P: narrows or widens the word choice pool.
• Repeat Penalty: available for Ollama, to reduce repetitive output.

Auto follows the selected model's recommended settings. If you want a stable baseline, start there. If you want the presence to take more risks or stay tighter, switch to one of the other presets and adjust the sliders.

Voice & Audio

This section controls how Ensoul speaks back to you and whether it adds ambient sound. It is separate from Voice Input, which is for talking into the chat box.

Text-to-Speech mode can be Off, System, ElevenLabs, or OpenAI. Use Auto-speak Responses if you want replies read aloud automatically instead of manually triggering them.

The ElevenLabs and OpenAI voice modes need their own API keys in Settings. Ambient sounds also rely on ElevenLabs and use API credits. If no key is set, the ambient toggle stays disabled.

Appearance

Appearance changes how the app looks and how it behaves visually.

• Skin: the standard clean look, or the Arcane skin when available.
• Theme schedule: Auto, Light, or Dark.
• Font size: adjust text scale if you want a tighter or roomier layout.

Arcane is a more ornamental skin. If you do not want the visual treatment, leave the default clean skin selected.

Experience

These toggles shape how much of the model's inner process and longer-term presence behavior you want to see.

• Show Thinking: reveals reasoning blocks when the model exposes them.
• Heartbeat: lets the presence think between sessions and sometimes reach out.
• Interaction Mode: Instrumental, Relational, or Immersive.

Interaction Mode changes the tone of the whole conversation. Instrumental is more clear and direct, Relational is warmer and more attuned, and Immersive leans into story, atmosphere, and character.

Changes here usually take full effect at the start of a new conversation.

About You

The About You section is a small profile about you that gets injected into conversations automatically. Use it for your name, pronouns, preferences, or any detail you want every presence to know without being told again.

The short context is the most important field. Fill that one first. The longer context gives larger models more room if you need to expand on the basics.

Tools

The Tools section is where you jump into the other major surfaces in the app. Each item opens a focused workflow.

• Summoning Portal: create, import, and manage custom presences.
• Memory Manager: inspect, edit, or delete memories per presence.
• Saved Moments: browse the messages and images you bookmarked.
• Image Gallery: view generated images and filter them by presence.
• Search All Conversations: search across all conversations and presences.

Backup & Restore

Full Backup & Restore saves your conversations, presences, memories, relationships, and settings to a single local file. It works across web, desktop, and mobile.

API keys are deliberately excluded from backup files. After a restore, paste your OpenRouter and voice-provider keys back into Settings.

The same section also includes Relationship Insights, which opens a deeper readout of mood, vitality, awareness, and related state, and Reset All Portraits, which clears custom portrait images back to the default glyphs.

Full Reset is the emergency option. It wipes all conversations, memories, companions, settings, API keys, and local state.

Safety & Reporting

Use Safety & Reporting when you need to report harmful content, a broken experience, or something that needs human review. The report form lets you pick a category, describe the issue, and optionally include a presence name or content ID so the team can find the right entry quickly.

If the built-in form fails, email support@ensoul.so directly.

Developer Mode

Developer Mode is an advanced troubleshooting surface. It is hidden behind an access password and is intended for testing prompt behavior and internal configuration changes.

Once unlocked, you can set a Session Override. That text gets appended to the system prompt for the current session, which makes it useful for short-lived experiments and prompt debugging.