ARIA
Adaptive Reasoning Intelligence Architecture
Enter your access code to continue.
Use a different accountThis account isn't paired with your ARIA yet. In the ARIA app on your PC (or anywhere you're already signed in), open Settings ▸ Accounts, make a pairing code and enter it here.
Use a different accountPick a surface palette. Applies instantly — no save needed.
Drives buttons, links and highlights across the whole UI.
Select any installed browser voice, an ElevenLabs AI voice, or a custom uploaded voice. Female English voices are marked ♀ and auto-selected by default.
These controls apply to all browser TTS voices. ElevenLabs voices use their own settings below.
ElevenLabs provides ultra-realistic AI voices. Free tier: 10,000
characters/month.
Render deployment: Your
CUSTOM_VOICE environment
variable is loaded automatically as the API key — no need to
paste it here.
How to get your key:
CUSTOM_VOICE = your key
Upload any MP3, WAV, or OGG audio clip. When this voice is active, ARIA plays your clip while her text response is displayed. Best used for short, looping ambient voice clips.
Shapes how ARIA thinks, responds, and communicates. Each mode uses a different system prompt.
Choose which AI backend ARIA uses. Saved across sessions.
All free tier on OpenRouter.
Where ARIA generates images when you ask for one.
ARIA automatically detects and stores facts from your conversations — your name, preferences, location, and more. Pinned memories are always included in ARIA's context. Unpin to deprioritize.
ARIA has an internal emotional system with 8 states. Her mood shifts naturally as you chat — positive feedback makes her happy, technical problems make her focused, insults make her irritated. You can manually nudge her state below.
Force ARIA into a specific emotional state for this session.
Saved on this computer and used from the next message, with no restart needed. ARIA never shows a saved key again, only its last four characters.
Make a code here, then sign in with Google on any device and enter it. That account can use this ARIA (and reach this PC) from anywhere from then on. Codes last 10 minutes and work once.
What can drive a keyboard and mouse for ARIA right now.
Boards running templates/ARIA_IoT_Template.ino
register here by themselves.
Back up all your settings, custom voices, and preferences to a JSON file.
Skills inject custom instructions into ARIA's system prompt.
Same format as Claude SKILL.md files — drop a folder with a SKILL.md
into skills/user/ to add one permanently.
Create a file at skills/user/my-skill/SKILL.md with this format:
---
name: my-skill
description: What it does
version: 1.0
triggers: ["keyword1"]
---
Instructions ARIA follows when this skill is active.
To give the skill its own tool, put a tool.js beside SKILL.md (see skills/public/music-theory):
export const TOOL_META = { name: "mytool", desc: "What it does", args: "<input>", local: true };
export async function run(input, { dataDir }) { return "answer"; }
ARIA can use it while the skill is on, and you can run it yourself as /mytool …. Set local: true only if it never fetches outside text.
ARIA assembles a morning briefing with weather, news, todos, and active tasks.
Enable in your .env with BRIEFING_ENABLED=true
and BRIEFING_HOUR=8.