There is a lot that you can do with a "build" process, and a bit of research with Google will no doubt fill in any gaps in your knowledge. For example, this Microsoft KB describes how you can control the number of times the autologon process will run before resetting itself back to normal logon:
http://support.microsoft.com/kb/221477
You need to set a default username and password, but this technique is ideal in automated build processes, as you can set up something in the HKLM RUNONCE key of the registry which will start a process next time the machine boots, using autologon parameters you have set. Your process can then put another entry in the RUNONCE key before rebooting the workstation, and another autologon process can then run yet another configuration process for you, before once again rebooting the workstation.
So you can have operations completed such as machine customisations, domain joining, installing management clients, defragging the hard disk - in fact whatever you need that cannot be incorporated in the base build for some reason.
In one of my past projects, we needed a single build that would support around 20 different desktop and laptop machines, and of course, each laptop had a long list of different utilities that were model specific, which therefore had to be installed after the image had been deployed to the PC. WMI was used to detect the exact model, and then a control program installed all required programs and utilities for that model, rebooted the machine, then completed tasks such as renaming the machine to its final computername, joining the domain, then removing the install account. As all the models we used supported asset tagging, we would code the machine name into the asset tag and read that during the post imaging processes, so no manual configuration was ever required.
Microsoft's Windows Automated Installation Kit and the MDT provide many of the tools to simplify build engineering, but there is always scope for additional fine tuning to meet individual requirements.