The kitchen had been on the list for two years. You know how it is. Every time we had someone over, we'd apologise for the worktop, the way one of the press doors hung slightly wrong, the extractor fan that sounded like a small helicopter. Eventually we stopped apologising and started actually planning.
The problem with planning a kitchen renovation, at least for us, was that we didn't really know how to start. We weren't sure what questions to ask a fitter. We didn't know what a realistic budget looked like. We didn't even know the right words for some of the things we wanted. So before we called anyone, we sat down with ChatGPT and spent an evening trying to get our thoughts in order.
We told it roughly what we had: a semi-detached house, a long narrow kitchen, a back door that couldn't be moved, existing plumbing we wanted to keep in place if possible, and a budget we were not going to mention publicly. We told it what we wanted: more worktop space, better storage, an island if it was feasible, and something that didn't look like it came free with the house.
What the AI actually got right
The most useful thing ChatGPT did was help us write a proper brief. Not a wish list. An actual document we could hand to a tradesperson without embarrassing ourselves. It asked us things we hadn't thought about, like whether we wanted soft-close hinges, how many people cook at the same time, what we store in the kitchen that we never use but can't throw out. Small questions, but they helped us figure out what we actually needed versus what we just thought we wanted.
It also gave us a rough framework for understanding costs. Not figures we'd quote anyone, more like a sense of proportion. It helped us understand why labour is often as significant as materials, and why getting a plumber and an electrician involved early matters rather than bolting them on at the end. We went into our first tradesperson meeting with at least a basic map of how a kitchen project tends to unfold.
Honestly, for someone starting from zero, this part was genuinely valuable. It's the kind of orientation you'd normally get from a friend who'd already done a renovation, and not everyone has that friend.
Where it started to show its limits
The AI had no idea about our local context, and that turned out to matter more than we expected.
When we eventually sat down with a local kitchen fitter, he looked at our brief, nodded at most of it, and then started pointing out things no chatbot could have told us. The wall we wanted to put shelving on had a slight damp issue that wasn't visible yet but would be a problem behind fitted units. The alcove we'd imagined turning into a pantry space was where the boiler flue ran. The island we'd been excited about was going to leave less than a metre of clearance on one side, which sounds fine until you've tried loading a dishwasher in that space.
None of that is ChatGPT's fault. It can't see your kitchen. It doesn't know that planning rules in certain counties are interpreted differently, or that certain suppliers have lead times that will push your whole timeline out. It doesn't know which local tradespeople are good or which materials hold up better in older Irish houses with less than perfect insulation.
There was also a moment where the AI gave us confident-sounding advice about building regulations that turned out to be slightly off for our specific situation. Not dangerously wrong, but wrong enough that we were glad we'd double-checked with someone who actually knew the rules here.
How we ended up using it
We stopped thinking of ChatGPT as an expert and started thinking of it as a very well-read assistant who had never actually set foot on a building site. It helped us prepare, organise our thinking, and ask better questions. The answers to those questions came from people who'd worked on real kitchens, in real houses, around here.
The brief we built with the AI's help did save time with the fitter. He said it was clear, which is not always the case. That alone was worth the evening we put into it.