How to use AI to cost a renovation project in France

I have spent the last 24 years thinking about converting our hayloft into living or rentable accommodation. When I say ‘thinking’, I really mean ‘dreaming’.

I have often looked at it from below and idly wondered how much it would cost to do it, and whether it would be worth it in the long run. My conjectures always petered out with me plucking a round figure out of the air that had no correspondence to reality.

Then, a few weeks ago, I decided I should do something practical about this possible DIY project. The only way I could think of for arriving at a budget was to make a long list of materials that I would have to buy. The more comprehensive and detailed the list, I reasoned, the more accurate the estimate.

Taking the first steps with AI

I was halfway through this self-imposed task when my wife (who knows little about DIY but a lot about time-saving common sense) asked me what I was doing at the top of a ladder fumbling around with a tape measure, pen and paper.

I told her and she replied with the killer question: “Why don’t you tell AI what you want to do and ask it how much it will cost?”

I spluttered about the value of hands-on human know-how versus an inhuman mathematical code without hands or experience of DIY shopping – and then admitted she may have a point. 

In the end, I decided to use both methods and compare the results, for your benefit, reader, as well as mine.

AI has some obvious advantages over the electric chatter of neurons and synapses but in this article and the next I am going to argue that the two approaches are complementary. There are still some good, rational reasons to be old-fashioned.

Make the system work for you

First, let me explain the project. I do not want to advertise any particular AI system, so let’s call the one I use Botch.

This is what I fed into its prompt box: “How much will it cost to convert a hayloft in a French farmhouse into living accommodation including two bedrooms, a bathroom with toilet and a kitchen. The space measures 7m by 10.7m and there is plenty of clearance overhead to make rooms of at least 2.5m high. The hayloft will need a 4m staircase to reach it.”

Whereas it took me about five days (not full-time, I stress), to come up with a figure for the materials I would need, it took Botch less than five seconds.

It assumed I wanted to pay professionals to do the work and returned an astronomical quote of €112,000 to €187,000 – or €1,500 to €2,500 per m². 

I refined my request to make it clear that I would do the work myself. That brought the price down significantly to €30,000 to €55,000 – or €400 to €750 per m² – but the first tranche of information was still useful because I might have to employ a tradesperson for the odd task. 

For a ballpark figure, with justifications provided, I was impressed. Naturally, Botch does not know everything about the specifics of my house and it asked me some questions so that it could hone its recommendations. I could go on forever tweaking the specs, but that was enough for me. 

Using it within a range of tools

The big plus point of using Botch – apart from the speed – is that it included items that did not immediately occur to me, such as the cost of an electricity compliance certificate and the tax implications for creating new living space.

Okay, I have to concede Round One to the algorithm; but in my next article, I am going to tell you why I still recommend that you make a low-tech, old-fashioned list yourself.