Getting Started

This guide helps you start using Ampere in a useful way.

OpenClaw is powerful, but self-hosting it means dealing with Node versions, install commands, provider keys, gateway setup, channel config, daemon/service setup, health checks, and config validation. Ampere handles that hard setup so you can start from the dashboard.

What Ampere Handles for You

If you self-host OpenClaw, the official setup includes steps like:

• installing OpenClaw on your machine
• checking Node.js compatibility
• choosing a model provider and adding API keys
• running CLI onboarding
• configuring the Gateway
• checking gateway status and ports
• setting up channels like Telegram or WhatsApp
• managing config files and validation errors
• keeping the service running

With Ampere, you do not need to start there. Your job is to use the assistant well: give it context, set useful automation, pick the right models, and improve memory over time.

Start with a Good First Prompt

Your first message should give the assistant direction.

Good first prompts:

• “Ask me 5 questions so you can understand how to help me.”
• “Help me set up your role as my assistant.”
• “Create a simple assistant profile for me.”
• “Explain what you can do in Ampere and help me pick my first useful task.”

Avoid starting with only “hi.” Ampere works best when your OpenClaw bot has real context.

Teach It How You Work

After the first prompt, tell the assistant things that should matter again.

Examples:

Remember that I prefer short answers with clear next actions.

Remember that my current focus is customer support and product growth.

When I ask for research, give me a short recommendation first, then details.

This improves memory and makes future conversations faster.

Check Heartbeat

Heartbeat is how your assistant can check in or notice important things on a schedule.

Use it for:

• personal productivity check-ins
• daily reminders
• lightweight follow-ups
• “tell me if something needs attention” workflows

Keep heartbeat instructions simple. If you make them too broad, the assistant may interrupt too often.

Set Up Useful Cron Jobs

Cron jobs are for recurring work.

Good first cron jobs:

• daily planning prompt
• weekly report summary
• morning task review
• website or analytics check
• reminder to follow up with leads

Start with one recurring job. Test it. Then add more.

Use Budget-Friendly Models

Not every task needs the most expensive model.

Use cheaper or faster models for:

• simple summaries
• rewriting text
• extraction
• classification
• quick chat
• basic research drafts

Use premium models when the task needs stronger reasoning, code quality, long context, or careful judgment.

See Credits & Pricing for model costs.

Save Credits and Cash

Simple ways to spend less:

• use shorter prompts when the task is simple
• attach only the files the assistant needs
• use budget-friendly models for routine work
• avoid asking the same long question many times
• create reusable skills for repeated workflows
• use BYOK if you prefer paying model providers directly
• keep cron jobs focused so they do not run unnecessary work
• disable features you are not using, such as extra cron jobs, broad heartbeat checks, or heavy tools, so they do not waste tokens in the background

Good setup means your assistant is useful without wasting credits.

Add Features Slowly

After dashboard chat feels useful, add more features one by one.

Common Questions