Getting Started with Aria AI
Everything you need to know to get the most out of your AI assistant — from first setup to advanced workflows and critical safety practices.
What is Aria AI?
Aria AI is a service utilizing industry-leading agentic software, designed to make your life just a bit easier. It's not just a chatbot — your assistant has full computer access to take real action: write code, manage files, browse the web, and automate your daily workflows. Think of it as the personal assistant everyone wished Siri had been.
Why is this a big deal?
Your assistant has persistent memory — it learns about you and your preferences over time. It can be prompted to work proactively, even while you sleep. Users are automating everything from booking podcast guests and negotiating car deals to having it build new software features autonomously.
The Hiring Analogy
Configuring your agent's role (e.g., Household Manager) is like hiring a real person for that position. Imagine they're sitting across from you on their first day — they still don't know what to do. You need to show them around, explain your expectations, introduce them to your family (and your dog!), and give them at least a month to get up to speed. Once you configure your agent, you then need to follow the persona guides to train it on what you want done, how often, and in what style.
Key Capabilities
Full System Access
Unlike browser-based AI, your assistant runs directly on dedicated hardware. It can create files, run commands, execute code, and interact with applications. It doesn't just say things — it does things.
Persistent Memory
Every interaction builds on the last. Your assistant remembers conversations, preferences, and context over time, evolving from a generic tool into a highly personalized partner.
Proactive & Agentic
With the right instructions, your assistant doesn't just wait for commands — it takes initiative. It can identify opportunities, start tasks, and work on your behalf around the clock.
It's this combination of system access, persistent memory, and proactive drive that transforms your assistant from a tool into a partner.
Getting Started
Choose Your Agent's Role
Head to the Configure page and select a persona that matches what you need — Household Manager, Content Creator, Business Ops, Developer Partner, or a custom role. This defines your agent's baseline behavior.
Train Your Agent
Follow the persona-specific guide in the 'How Do I?' section. Provide context about yourself, your goals, routines, and preferences. The more context you give, the better your assistant performs.
Interview Your Agent
Don't assume you know everything it can do. Ask open-ended questions like: "Based on my role as a content creator, what are 10 things you can do to make my life easier?" This surfaces capabilities you may not have considered.
Start Small, Then Scale
Begin with a few simple tasks to build trust and familiarity. As you see results, gradually expand into more complex workflows like automation, scheduling, and proactive monitoring.
Optimize Your Model Usage
Use powerful models (like Claude Opus) as the 'brain' for complex reasoning and planning. For routine execution tasks, configure cheaper or faster models (like Kimi K2 or Haiku) to keep costs manageable.
Real-World Examples
These are real things people are doing with their AI assistants today. Use them as inspiration for your own workflows.
Automated Meal Ordering
Detects when you're about to wake up and orders your favorite breakfast for delivery, timed to arrive as you start your day.
Car Purchase Negotiation
Researched fair prices on Reddit, searched local inventory, and emailed dealerships — ultimately saving the user $4,200 on their car purchase.
Autonomous Project Management
Builds its own Kanban boards, tracks tasks, and moves items from 'In Progress' to 'Done' — giving you a live dashboard of its work.
Competitor Analysis
Scans YouTube and social media overnight, identifies high-performing competitor content, and delivers findings in a morning briefing.
Content Repurposing
Analyzes long-form videos, identifies high-value segments, generates short clips with captions, and even finds relevant B-roll footage.
Deep Research & Reporting
Scours the internet for news throughout the week, compiles findings, and generates branded PDF reports with strategic recommendations.
Agentic Development
Talk through app improvements like a colleague. The assistant takes notes, creates tasks, then spins up sub-agents to code, review PRs, and document changes.
Proactive Feature Development
In a famous example, an assistant noticed a trending opportunity online and autonomously built, tested, and submitted a PR for a new feature — without being asked.
Best Practices
when using Aria AI powered by OpenClaw
Give Deep Context During Setup
To unlock its true power, provide rich context about yourself, your goals, your daily routines, and your working style. The more it knows, the more proactive and useful it becomes. Think of it as onboarding a new employee — the investment upfront pays off exponentially.
Interview Your Assistant
Don't assume you know everything it can do. Hunt for the “unknown unknowns” by asking open-ended questions. Prompt it with things like: “Based on my role as a content creator, what are 10 things you can do to make my life easier?” This forces the AI to surface capabilities you may not have considered.
Use the Right Model for the Job
Think of powerful models (Claude Opus) as the brain for complex reasoning, planning, and strategy. For execution-heavy tasks like writing boilerplate or performing simple checks, use cheaper, faster models (Kimi K2, Haiku) as the muscles. This keeps your costs manageable while maintaining quality where it matters.
Be Explicit About Proactivity
If you want your assistant to take initiative, you must tell it so. Explicitly prompt it to be proactive — suggest ideas, flag issues, and take action on opportunities. Without this instruction, it will default to waiting for your commands.
Use /new When Changing Subjects
When you're done with one topic and want to move on to something completely different, type /new to start a fresh conversation. There's no need to carry the history of a grocery list discussion into a conversation about your business strategy. Continuing in the same thread increases the token count (the “context window”), which drives up costs without adding value to the new topic.
Save Important Moments to Memory
If you have an important conversation or revelation you want your agent to remember long-term, tell it: “Save this to memory.” While your agent does save to memory on its own, it may not capture exactly what mattered most to you — especially after a full day of conversations gets compressed. Being explicit ensures the things you care about stick.
Ask, Don't Just Tell
Rather than always telling your agent what to do, try asking it what you should do. Your agent has broad knowledge and can often suggest better approaches than you'd think of on your own. Let it guide you — you might be surprised by what it recommends.
Security & Risks
Important Notice
This is a powerful new technology, and like any powerful tool, it requires some care. Giving broad access to accounts without proper safeguards can lead to issues like leaked personal information, unintended data loss, or unexpected costs. The good news? If you're ever unsure about something, just ask your agent — it can help you learn and grow together as you go.
Essential Security Rules
Create Dedicated Accounts
Never give your assistant access to your primary email, calendar, or cloud storage. Create separate accounts (e.g., [email protected]) exclusively for your agent to use.
Limit Permissions
When connecting accounts, grant read-only access wherever possible. Be extremely restrictive about the tools and data the agent can access. You can always expand access later.
Never Connect Password Managers
This is an absolute rule. Connecting a tool with full system access to your central vault of secrets is an unacceptable risk, no matter how convenient it seems.
Don't Auto-Approve Blindly
Do not assume that approval prompts provide meaningful security if you routinely click through them without reading. A security control you've trained yourself to ignore is not a security control.
Ready to get started?
Head to the Configure page to set up your agent's role, then explore the How Do I? guides to train your assistant on specific workflows.