• Besides the fact that you can’t use workers, are there key differences between creating a deployment and submitting a run of that deployment via API vs running locally still pointing to a remote PREFECT server installation? • I remembered prefect 1 had a GraphQL API, are we planning to discuss the API later? is there a reason why the Prefect 2 doesn’t have continued on the GraphQL path?

1. If I understand your question, using the deployment instead of running as a Python script gives you the ability to observe a lot more about your workflows, schedule them, and control more aspects of them. 2. The GraphQL API in Prefect 1 created a number of issues - it was found to be brittle and difficult to maintain. Prefect 2 won’t have a GraphQL API.

Can you provide more detail about (1). What more can you observe, for example, and what are the additional aspects you can control?

If you interact with the flow from the UI, the deployment is the way to go. That deployment contains a lot of info - you can see all of it in the UI. You can control the infrastructure a flow runs on through deployments - via work pools and workers. You can look at task run concurrency in the UI. You can work with notifications.

If you interact with the flow via API, like if you want to have an internal frontend, probably the deployment is still the way to go?

Or maybe the REST API: https://docs.prefect.io/latest/api-ref/rest-api/