1. Store Your API Keys in the Vault
Your agent can connect to external services like Google, GitHub, or any REST API. But those services need API keys, and you do not want those keys sitting in plaintext.
OmniClaw includes an encrypted credential vault that stores your keys outside the sandbox. They are injected at runtime and never touch the filesystem. To add keys, open your dashboard and navigate to the vault section.
Common keys to store: OpenAI API key (if you want to use your own), Google Calendar credentials, GitHub personal access token, weather API key, or any service your agent's skills need.
2. Try Different Prompts
Most people send "hello" and stop there. Your agent can do much more. Here are 10 prompts to try right now:
Daily productivity
"Summarize the top 5 tech news stories from today"
Writing assistant
"Rewrite this paragraph to sound more professional: [paste text]"
Learning
"Explain quantum computing like I am 12 years old"
Travel planning
"Plan a 3-day trip to Lisbon with a $150/day budget"
Coding help
"Write a Python function that converts CSV to JSON"
Health and fitness
"Create a 4-week beginner workout plan I can do at home"
Language practice
"Have a conversation with me in Spanish, correct my mistakes"
Meal planning
"Suggest 5 dinner recipes using chicken, rice, and broccoli"
Decision making
"Help me decide between buying and leasing a car. Ask me questions first."
Creative writing
"Write a short bedtime story about a cat who learns to fly"
The key is to be specific. Instead of "help me with cooking," give your agent context: what ingredients you have, how much time, dietary restrictions. The more detail you provide, the better the response.
3. Choose a Different AI Model
OmniClaw comes with a default model, but you are not locked in. The built-in LLM proxy supports multiple providers and models. Each has different strengths:
- GPT-4o — Fast, good at general tasks, strong at following instructions
- Claude Sonnet — Excellent at nuanced conversation, long-form writing, and analysis
- Gemini Pro — Strong at reasoning and multilingual tasks
- GPT-4o Mini — Fastest and cheapest, great for quick answers
Browse all available models and their capabilities on the models page. You can switch models at any time from your dashboard without losing conversation history.
4. Set Up a Second Channel
Your agent does not have to live on just one platform. OmniClaw supports running WhatsApp and Telegram simultaneously on the same instance. Same agent, same skills, same conversation context — accessible from both apps.
Why run both? Some practical reasons:
- Use WhatsApp for personal conversations and Telegram for group chats or shared bots
- Keep a backup channel in case one platform has an outage
- Let different people reach the same agent on their preferred platform
To add a second channel, go to your dashboard and click Add Channel. The connection process is the same as when you first set up — scan a QR code for WhatsApp or paste a BotFather token for Telegram.
5. Share with Friends
The best part of running an AI agent on a messaging platform is that other people can use it too. Here is how sharing works on each channel:
Your agent is linked to your phone number. Others can message it by messaging you — the agent responds on your behalf when it detects a prompt. You can also add your agent to a WhatsApp group, where it will respond to messages directed at it. Just create a group, add your phone, and tell friends to join.
Telegram
Your Telegram bot has a public username. Anyone can find it by searching for @YourBotName in Telegram. Share the link t.me/YourBotName with friends and they can start chatting immediately. You can also add the bot to group chats where it will respond when mentioned or when a message starts with a slash command.
Usage from friends counts toward your LLM credit. Keep an eye on your dashboard to monitor spend. OmniClaw tracks usage per-message so you always know where your credit is going.
Not connected yet?
Deploy your own OpenClaw agent in under 2 minutes. $5 free credit, no card required.
Get started