Home Assistant occupies an awkward place in a professional installer's toolkit. It will do things Control4 and Savant simply cannot, it costs nothing, and it runs on hardware you could hide behind a book. It is also the platform most likely to leave a client stranded when the person who configured it moves on.
Both things are true. Which one matters depends on the house.
What it is genuinely better at
Integrating things nobody else will. Every closed platform has a list of supported devices, and if your heat pump, your EV charger or your pool controller is not on it, you are waiting for a manufacturer to care. Home Assistant has thousands of integrations and a community that writes new ones in days. When a client has already bought something unusual, this is often the only route that does not involve replacing it.
Automation logic with actual depth. Most control systems give you scenes and simple triggers. Real automation needs state, timers that survive a reboot, conditions that consider three things at once, and the ability to say "unless somebody is already in that room". Writing that in Home Assistant is straightforward. Writing it in a keypad-driven system usually is not.
Local by default. No cloud account is required for the core system to work. In a house where the broadband goes down for two days after a storm, that distinction stops being philosophical.
Cost. The software is free and the hardware is inexpensive. On a project where the budget has gone to lighting and joinery, it can be the difference between having automation and not.
What it is genuinely worse at
It expects maintenance. Updates land monthly and occasionally break integrations. Something that worked in April may need attention in June. A client who is happy to tinker will not mind. A client who wants the house to be finished will mind a great deal, and will be right to.
The interface is the weak point. The dashboards are configurable rather than designed, and it shows. If the family interface matters — and in a high-end house it usually matters more than the automation itself — you are either doing significant design work or pairing it with proper keypads.
Nobody is accountable. With a closed platform there is a manufacturer, a dealer channel and a support path. With Home Assistant there is a forum. That is fine while you are the integrator; it is a problem the day a client calls someone else.
No commissioning discipline. Because it is so easy to add things, systems accrete. We have inherited installations with four hundred entities, half of them from devices that left the house years ago.
How we actually decide
The question we ask is not "which platform is better", it is who is going to look after this in three years.
If the answer is us, on a care plan, Home Assistant is often an excellent choice and we will happily build on it. Documented, version-controlled, with updates applied on our schedule rather than automatically, it is stable and immensely capable.
If the answer is "nobody, really", we specify something with a dealer network behind it, even where that means accepting a less clever system. A house that does something simple reliably for a decade beats one that did something brilliant for eighteen months.
Where we tend to land
The most robust systems we install are frequently hybrids. Lighting on a wired, closed, boringly reliable platform — Lutron or KNX — because lighting must work when everything else is broken, and because a light switch is not a place to be clever. Then Home Assistant sitting above it as the integration and intelligence layer, tying in the heat pump, the EV charger, the cameras and the AI, with keypads and panels as the interface the family actually touches.
That way the failure mode of the clever part is that the house becomes merely a normal house, rather than a dark one.