Deploy and run your AI apps and agents.
Run the application or agent beyond your machine, with its configuration, the data it keeps, and visibility into how it operates.
project: inbox-agentprocesses: worker: command: ["python", "agent.py"]A research agent, triggered through an API.
Or a document-processing app, or a background task that calls a model.
- 1
Deploy the app or agent
As a web process, an API, or a worker.
- 2
Configure model access
Keep API keys and settings as secrets with the project.
- 3
Watch it work
Follow the logs as it runs, and its data where it keeps it.
What it can use
- Apps and workers. Run the agent as a web process or in the background.
- Configuration and secrets. Model keys and settings, kept with the project.
- Data resources. PostgreSQL, Valkey, and volumes for what it keeps.
- Deployment visibility. Logs, releases, and status as it runs.
$ shpyrd deploy==> Archiving HEAD (654f4925638e)==> Uploading source (2.6 KiB)==> Building===> detect4 of 9 buildpacks participating===> export==> Releasing Running: web 3/3 · worker 1/1Released v3: Deploy 654f4925638ehttps://shop.acme.shpyrd.appPractical questions
Does Shpyrd run the model?
No. Your application calls the external model service you choose; Shpyrd runs the application.
Who defines how the agent behaves?
Your framework and your code. Shpyrd gives it a place to run, its configuration, and visibility into its operation.
Can it run untrusted code from users?
Running arbitrary untrusted code is not part of what Shpyrd offers today.
Run it beyond your machine.
Deploy your AI app or agent with what it needs to keep running.