Skip to main content
Once you’ve built and tested your agent locally, you can deploy it to Blink Cloud to make it accessible to others and unlock powerful debugging features. 
blink deploy
This command packages your agent and deploys it to Blink Cloud, where it runs in a managed environment.
Blink agents never require the cloud to run. Deployment to Blink Cloud is completely optional. Blink is MIT licensed and will always support local-first development.
Deploying to Blink Cloud provides several benefits beyond local development:

​
Multi-User Access

Once deployed, your agent becomes accessible to other users. Share your agent with your team, customers, or the broader community.
  • Web interface: Users can chat with your agent at blink.so
  • Slack integration: Deploy your agent to Slack channels
  • GitHub integration: Use your agent to respond to issues and pull requests
  • API access: Integrate your agent into custom applications

​
Deep Logging and Tracing

Blink Cloud automatically captures detailed logs and traces for every interaction with your deployed agent:
  • Message history: Full record of all conversations and agent responses
  • Tool execution logs: See which tools were called, with what inputs, and what outputs they produced
  • Token usage tracking: Monitor model token consumption across all chats
  • Performance metrics: Response times and processing duration for each interaction
Logging Tracing

​
Usage Metrics

The Usage screen provides real-time metrics for your agent over the last 30 days, helping you understand how your agent is being used and performing:
  • Total runtime: Cumulative processing time across all conversations
  • Total chats: Number of conversations initiated with your agent
  • Tokens processed: Total token usage across all model interactions
  • Time to first token: Average latency before the agent starts responding
  • Daily runtime: Day-by-day breakdown of processing time
  • Daily chats: Conversation volume trends over time
  • Token usage over time: Track token consumption patterns
  • Response time by model: Compare performance across different LLM models
These metrics help you optimize your agent’s performance, manage costs, and understand user engagement patterns. Usage

​
Deployment Workflow

1

Test locally

Use blink dev to build and test your agent thoroughly before deploying.
2

Deploy to cloud

Run blink deploy to package and upload your agent to Blink Cloud.
3

Monitor and debug

Use the Blink Cloud dashboard to watch your agent in action, review logs, and measure performance.
4

Iterate

Make changes locally, test with blink dev, then redeploy with blink deploy to push updates.

​
Managing Deployments

Each time you run blink deploy, Blink creates a new deployment with an incremental version number. This allows you to:
  • Roll back: Revert to previous versions if issues arise
  • Compare versions: See what changed between deployments
  • Test in production: Deploy experimental versions alongside stable ones
View all deployments for your agent in the Blink Cloud dashboard under the Deployments tab. Deployments Deployment Details

​
Environment Variables

Your deployed agent can access environment variables you configure in the Blink Cloud dashboard:
  1. Navigate to your agent’s Settings page
  2. Go to the Environment section
  3. Add your API keys, secrets, and configuration values
These variables are securely stored and injected into your agent at runtime. Never commit secrets to your code - use environment variables instead.
Only variables set in your .env.production file are deployed with your agent. Variables in .env.local are for local development only.
⌘I