How to smartly wire door/window contacts/buttons/LEDs? HA vs Dali-2 vs KNX etc.

I haven’t read everything in detail, but in summary, I think it is about preparing a decentralized KNX system by expanding the "normal" electrical installation. Initially, this is probably the most expensive option, but the costs are not incurred all at once.

I would do the following: Convince the provider to wire the rooms not with 3-core but with 5-core cables. This way, you can already have at least two centrally switched devices per room (by connecting them to the switched lines). If he is reasonable, you can have two cables laid at once, allowing for 5 independent switches/dimmers distributed in the room. You can easily accommodate the blinds and window contacts decentrally in the respective reveal. Besides the KNX cable, you only need to drill the empty socket for this. KNX light switches are typically installed higher than the classic ones (due to the display), so you should immediately install the socket with the KNX cable at the appropriate height.

This covers most cases. There are also decentralized actuators that fit in a deeper flush-mounted box (for example, behind the normal socket). Basically, you should deal with how KNX can be installed during a renovation. You would have the advantage that you don’t have to plaster twice. Additionally, you need to reserve space in the technical room.

Thanks again!
I now understand what you mean, I can only use KNX very limitedly in my case, and due to the decentralized operation (since I can't just run direct 230V lines to every actuator) it becomes significantly more expensive because I need actuators everywhere instead of using a few central actuators in the basement that can control many devices, e.g., one shutter actuator for 4-8 shutters.
Originally, I just wanted to misuse KNX/Dali as a wired "alarm system" and collect the door/window contacts "intelligently / with less wiring effort over BUS."
The problem is the costs anyway; the builder wants 200 euros for every single line he additionally pulls from the basement to an arbitrary spot (empty box), so if I let him pull BUS to just one single box in 5 rooms, that already adds up to 1,200 euros. But I had already budgeted for that. And with the remaining wiring (whether to the switch or the actuator) done by myself, not much more will be added, just material costs (and hardware). Therefore, my idea was that this solution is still cheaper than if I lay a separate 4-core/8-core bell wire for each door/window contact into the basement + the advantage that the switches are then already on the BUS and don't have to be wireless (Zigbee). I wanted to replace some existing switches with push-buttons anyway at some spots, which can then control certain lamps/scenarios via HomeAssistant.

My FINAL idea: KNX ONLY for shutters/binary contacts with KNX flush-mounted shutter actuators with binary inputs, no lights, no push-buttons; I will have the BUS come out under the main shutter switch (which has to be separated anyway, either a separate box or a double box with mechanical separation), this BUS then goes into the basement, power supply (100€), connected to HomeAssistant via IP interface (150€). Actuators cost 50-80 euros (other actuators like Rademacher, Shelly, etc. also cost something, sometimes just as much as I wanted to use anyway). Then I would be at 890 euros for 8 roller shutters + 1,200 euros for laying cables = total 2,100 euros.
And if I feel like it later, when the positions of other components are fixed, I can still lay BUS cables to these components flush-mounted. Everything else otherwise over Zigbee and Shellys.

Or do you also consider this solution a mess for my specific situation? A non-KNX solution is pretty much just as expensive (mainly because of the expensive installation costs).

I could also have these BUS cables placed in a larger empty conduit and still have 1-2 additional 8-core cables laid with them just in case; whatever you have, you have.

KNX-RF can also be an option in this constellation. Again, of course it is somehow feasible, the critical factors are time and money. There is already no time, and the TE has to decide how much money it is worth to him. I still bet that in the end, after spending a lot of money in several places, the right cables are not there and one either has to settle for compromises or move from one improvised emergency solution to the next.

What about bus cables in the ceiling? In the exterior wall? Outside (garage/carport/weather station/terrace/...)?

What type of motion/presence detectors do you plan to install (this affects the suitable installation location)?

Do the brands you have selected support the required functions?

Have you planned a central location for an RF receiver and is the radio range from this location sufficient to reach all necessary areas?

My FINAL idea: KNX ONLY for blinds/binary contacts with KNX flush-mounted blind actuators with binary inputs, no lights, no switches,

And what are you going to do with that? Compared to a classic installation, you can then freely control blind groups, nothing more. For that, KNX is overkill.

no switches

How do you then operate the blinds?

connected via IP interface (150€) to HomeAssistant

What does HA do then? Control the blinds? With what data?

The problem is the costs anyway, the house builder wants 200 euros for every single cable that he additionally pulls from the basement to any random point (empty conduit)

That is just a polite way of saying: "I don't want to do that!"

Or do you also find this solution to be a mess for my specific situation? A non-KNX solution is pretty much just as expensive

Yes! It’s a mess. You don’t have a special situation either – except that you have high but normal costs for special features because of the construction contractor (wrongly chosen for your needs). Either "properly," then one day is no longer enough for planning, or completely different (I’m a fan of Shellys and iOBroker).

How about bus cables in the ceiling? In the exterior wall? Outside (garage/carport/weather station/terrace/...)?
Nothing planned in the ceiling, since I will probably mainly use Paulmann ZigBee LED strips anyway (2nd recessed ceiling with space for drivers/controllers)
Only surveillance cameras (UniFi) planned on the exterior wall via Cat7 PoE to UniFi gateway/switch
Only power + power for wallbox + Cat7 planned to the garage. Terrace not planned extensively, just power, access point facing the garden, camera, sensors simply via ZigBee.

What type of motion/presence detectors do you plan to install (this affects the suitable mounting location)?
Senvolon motion detectors (Zigbee) powered via USB-C, said to be very good

Have you planned a central location for an RF receiver, and is the radio range from there sufficient for all required areas?
I could place it well in the stairwell where there is no concrete ceiling, it reaches up to the attic.

And what do you do with that? Compared to a classic installation, you can then freely control blind groups, nothing more. For that, KNX is overkill.
Mainly just to somehow wire and read all door-window contacts and glass break detectors so that I can further process them in HomeAssistant, trigger scenarios, LEDs, sound, alarms, notifications. For example, raise blinds halfway when a window opens, notify when leaving the house that a window is open, or when sitting on the terrace that blinds do not go down.
That blinds are then also operated via a KNX actuator is rather incidental, otherwise it would happen via Shelly/Rademacher actuators or similar.

How do you then operate the blinds? What does HA do? Control blinds? With what data?
Via switches (Zigbee) and voice command (Home Assistant Voice and Alexa)

--
As mentioned, the KNX solution is probably about the same price as without KNX.
By gut feeling, I somehow feel safer if at least such "security contacts" a) are wired and b) KNX might be more fail-safe than a solution via Homeassistant ESPHome, or is that relatively irrelevant? And wireless, KNX RF/Zigbee/WLAN is certainly manipulable if a burglar comes with a jammer?