Skip to main content

​
Overview

Deploy your agents with ziet deploy. Your code goes live with API endpoints, CLI access, and dashboard UI. 
ziet deploy
That’s it. Your agent is live at https://api.ziet.ai/agents/{agent_id}/run

​
Quick Deploy

1

Write your agent

# agent.py
from ziet import Agent, Action, memory

@Action(
    id="process_data",
    name="Process Data",
    description="Process input data"
)
def process_data(data: str) -> None:
    result = data.upper()
    memory.add(key="result", value=result)

@Agent(
    id="my_agent",
    name="MyAgent",
    instructions="Process data using the process_data action",
    actions=["process_data"]
)
class MyAgent:
    pass
2

Deploy

ziet deploy
Output:
🚀 Deploying to Ziet...

✓ Analyzing code...
✓ Building agent...
✓ Deploying my_agent...

✅ Deployed successfully!

Agent: my_agent
Endpoint: https://api.ziet.ai/agents/my_agent/run
Dashboard: https://dashboard.ziet.ai/agents/my_agent
3

Test it

ziet run my_agent --data "hello"

​
Deploy Commands

​
Deploy All Agents

Deploy all agents in your project: 
ziet deploy

​
Deploy Specific Agent

Deploy only one agent: 
ziet deploy my_agent

​
Deploy to Environment

Specify production or development: 
ziet deploy --env production
ziet deploy --env development

​
Deploy Options

# Dry run (validate without deploying)
ziet deploy --dry-run

# Force redeploy (bypass cache)
ziet deploy --force

# Verbose output
ziet deploy --verbose

​
Project Structure

Recommended layout: 
my-agent/
├── agent.py              # Agent definition
├── actions/              # Action functions (optional)
│   ├── __init__.py
│   ├── search.py
│   └── process.py
├── requirements.txt      # Python dependencies
└── .env                  # Local env vars (git-ignored)
Keep it simple - all you need is your Python file and requirements.txt.

​
Configuration (Optional)

​
requirements.txt

List your Python dependencies: 
requests==2.31.0
openai==1.3.0
stripe==7.0.0
Ziet automatically installs these when deploying.

​
Environment Variables

Set secrets in the dashboard:
  • Dashboard
  • CLI
  1. Go to Settings → Environment Variables
  2. Click Add Variable
  3. Enter key and value
  4. Click Save
  5. Deploy (variables are automatically available)
Access in code: 
import os

api_key = os.getenv("OPENAI_API_KEY")

​
Environments

​
Development vs Production

Test in development before going to production: 
# Deploy to development
ziet deploy --env development

# Test it
ziet run my_agent --env development --data "test"

# Deploy to production
ziet deploy --env production

​
Versioning

​
Automatic Versioning

Each deployment creates a new version: 
ziet deploy

# Output: Deployed my_agent@v23

​
Rollback

Roll back to a previous version: 
# List versions
ziet versions my_agent

# Rollback
ziet rollback my_agent --version v22

​
Best Practices

Never hardcode API keys:
# ✅ Good
api_key = os.getenv("API_KEY")

# ❌ Bad
api_key = "sk-12345..."

# Test locally
ziet run --local --agent my_agent

# Then deploy
ziet deploy

# Test in dev
ziet deploy --env development

# Promote to production
ziet deploy --env production
Only include what you need in requirements.txt:
# Only list packages you actually import
openai==1.3.0
stripe==7.0.0

​
Troubleshooting

Solutions:
# Check for syntax errors
python agent.py

# Validate deployment
ziet deploy --dry-run

# Check logs
ziet deploy --verbose
Solution: Add to requirements.txt
# requirements.txt
requests==2.31.0
openai==1.3.0
Then redeploy:
ziet deploy --force
Solutions:
  1. Check dashboard: Settings → Environment Variables
  2. Verify exact name:
key = os.getenv("API_KEY")  # Must match dashboard
  1. Redeploy:
ziet deploy --force

​
Next Steps

Invoking

Run your deployed agents

Monitoring

View logs and metrics

API Reference

REST API documentation

Agents

Learn more about agents