The platform is the one decision that constrains every decision after it. Pick a closed ecosystem and your device choices narrow permanently; pick an open one and the work moves to you or to whoever you hire. Here is the honest comparison.
The platform constrains every decision after it. Closed ecosystems like Apple Home, Google Home and Alexa start easier and hit a ceiling sooner. Home Assistant has the highest ceiling and the steepest setup, and that trade is the whole argument. You can run it and still use Siri, Alexa or Google as the voice layer.
The short version
The cleanest interface and the best privacy posture of the mainstream platforms. Automations run on a home hub in your house, so a lot of it is genuinely local. The cost is device choice: the certified device list is narrow, particularly for aircon, curtains and anything sold locally in Singapore, and automation logic is deliberately simple. Excellent as a control surface, limiting as a foundation.
The best voice recognition and the most forgiving setup. Almost everything works, right up until you want it to do something conditional. Automations are cloud-dependent, which means your lights depend on a data centre being reachable. Google has also retired products and features on its own schedule more than once, which is a real risk to build a house on.
The widest device support of the voice platforms and a large skills ecosystem. Same structural issue as Google: the logic runs in the cloud, the experience depends on your internet, and the feature set moves when Amazon decides it should.
The most capable of the mainstream hubs, with genuine Zigbee, Z-Wave and Matter support in the hardware. It sits between the closed ecosystems and Home Assistant. The catch is that the platform has been rearchitected more than once, taking working setups with it, and the local execution story is partial rather than complete.
Open source, runs on hardware in your home, and integrates with well over two thousand systems including nearly everything sold in Singapore. Automations are as complex as you want, data stays on your storage, and there is no subscription and no vendor who can change the terms. The cost is that it is a system rather than a product: it has to be set up properly, maintained, and backed up. That is exactly the work we exist to do.

Not "which is best" - "what happens when the vendor changes their mind?" Every closed platform has, at some point, moved a feature behind a subscription, discontinued a hub, or dropped support for a device category. If that would leave you with hardware in your walls that no longer does what you bought it for, you have made an infrastructure decision on a consumer product.
The second question is whether you will ever want a condition. Not "turn on the light" but "turn on the light at forty percent, only after sunset, only if someone is home, and not if the TV is playing". Closed platforms handle the first and struggle with the second.
This is the part people miss. Home Assistant exposes your devices to Apple Home, Google Home and Alexa, so you can run everything locally and still say "Hey Siri, goodnight". The voice assistant becomes a control surface on top of your own system rather than the system itself.
That is the configuration we build most often: local logic, local data, and whichever voice assistant the household already talks to.
Home Assistant is only as reliable as its hardware, and this is where most self-built systems fail.
You can move from a closed ecosystem to Home Assistant without throwing hardware away. Most devices that work with Apple Home, Google or Alexa also work with Home Assistant, often more completely, because the integration talks to the device rather than to the vendor cloud. The devices you lose are the ones that were only ever a rental.
A complimentary design session covers all of this against your own floor plan, with a live Home Assistant system in front of you.