Currently it appears that when the SCCM agent in the WinPE boot image connects to a Management Point, it selects one without using the Boundary Group rules, instead selecting one from the AD published list, starting with the first alphabetically.
Instead this should behave like the full blown Windows SCCM Client and select a Management Point based on Boundary Groups such as IP range.
It is worth noting that Distribution Points are selected correctly according to Boundary Group by the WinPE client, for OSD content.

Sometimes it might be handy to remove a computer from the OSD collection as a final step during the task sequence. Or additionally add the computer to additional collection(s) for for example extra software deployment or another task sequence.
Now scripts need to be created therefore and in my experience this doesn't always work well.
A predefined step where the collectionname or collectionID needs to be added to add (or remove) the computer to (or from) would be much appreciated.

Could you add Revision History for task sequences? It would save you the effort of exporting and importing task sequences or keeping copies for References. This would be a nice to have feature in the Configuration Manager Console - although I'm sure we all have a scheduled job for exporting SCCM configuration automatically anyway.

It would great to be able to configure TFTPBlocksize and TFTPWindowsSize on PXE-enabled DPs from the Configuraiton Manager console, and not having to edit the registry or to use Configuration Item/Baseline to do this. An option on the PXE tab in the Distribution Point properties please?!

Maik Koster has created an excellent Web Service for performing tasks e.g. during OSD, like removing a device from a collection. I think it would be a great idea to have a Site System Role for OSD where you have a built-in step in the TS to choose various tasks to perform during OSD.

Provide an option to RESUME an already running OSD task sequence deployment that failed in a specific area versus having to start the deployment all over again. In the event of a failure, this would allow for us to resolve the issue right away and then allow the admin to RESUME the OSD task sequence deployment from the exact point of where it failed (not have to start everything over again).

Give us the option to restrict a required Task Sequence to "When no user is logged on" This would be particularly helpful when using Task Sequences to push out Application updates. That way if the TS tries to run while a user is logged on it won't fail as soon as it hits the applications which are set to "When no user is logged on"

We are seeing a number of unexpected restarts during the "Upgrade Operating System" Windows 10 feature update that causes the task sequence to fail. These restarts are happening during the "setup.exe" phase of the process. Allowing the task sequence to continue would resolve this issue.

It would be cool, if you set the CCMEXEC Service to "Auto" automatically while running a Task Sequence. After OS Deployment run, it sould be set to "Delayed" again.

This would Speed up OS Deplyoment up to 15 Minutes!

(We tested it in here with a TS Step after CCMSetup Step) cmd /c sc config "CcmExec" start= auto

Sometimes a task sequence can fail without the technican noticing the reboot countdown, so they just see a login screen and assume everything is okay. How about a "OS Deployment Failed/Succeeded" splash screen at completion? MDT (UDI) has a completion screen, although it doesn't show on error. Still - something like that.

Right now GPO processing is disabled during TS execution. This is annoying as quite often there's settings and environment configuration for a machine pushed by GPO, and right now they only apply at the end of the TS.

It would be nice to have an explicit "process group policy" step in a TS, which would be the equivalent of a "gpupdate /force".

It would be good to be able to add the latest dynamic updates to a "Operating System Upgrade Package" to ensure it have the latest compatibility database and setup files.
Our environment do not allow the setup process to access internet + even if it could it will be better if we can download once to the package and not having all clients going out to internet.

We use in Switzerland a EN-US MUI Windows 7 (DE/FR/IT) an we plan inplace upgrade to Windows 10 1607. After Upgrade Windows 10 is only in english upgraded. (We can only use a en-us win10, another language image crashed the TS) Please add a step for set one or more LP in the upgrade process (Native, please no Scripting, MDT etc.)
The same Procedure we require for the Win10 Upgrades via WSUS (1511->1607->1703...) After upgrade we must add manually via Tasksequence the other language packs. Why?

Currently (v1806 or later) Third Party Software Updates are not being installed during the Install Software Updates step in a Task Sequence.

Ideally third party software updates should be installed just like regular updates during the Task Sequence. A machine should be fully patched (secure and end-user ready) when it leaves the deployment bench.

Please add to the Restart Computer OSD task sequence step an option that will check if the system requires a reboot and then only restart the computer if it needs to. Right now to work around this need, we have to execute a PowerShell script and write the results to a custom task sequence variable and then use that variable as a condition on the Restart Computer step. Adding this option natively would be much more elegant. Attached is the script we're using.