This was more of an intro to show that you don't need to do much coding to create a web service to access relational data.

I have just started using this technology mostly because I want to move toward web services.  I played with web services in the past trying to serialize data for client side Javascript (JSON), which was difficult at best (it was easier to code it by hand).  This seems useful in that you can create a diagram (edmx ADO.NET Entity Data Model) to represent your tables and relations, point this web service to it (svc ADO.NET Data Service), and bingo, instant serialized access to all the data.

Then, you can use Atlas (Microsoft Ajax) to do CRUD (create retrieve update and delete) operations within Javascript at the client browser without knowing a lot of what you are doing.  Of course, if you get to this point, you probably know how to do these things anyway.  This means you can create a Javascript variable, point it to the database "context" (or in this video, the "service:"), and have array access to records and object access to fields.

It's pretty cool stuff when you don't want all the details of coding, but isn't that the point of abstracting these technologies anyway?