The first cold snap of October arrived on a Sunday this year, and by Monday morning my living room had that particular chill — the kind that sits in your knees and whispers about the heating bill. I stood in the hallway in a jumper, staring at the thermostat like it owed me money, and thought: there has to be a smarter way to do this than panic-googling “why is my house cold” every autumn.
There is. I spent the last two weeks treating my flat like a project and ChatGPT, Claude, and Gemini like the crew — energy auditor, draught detective, budget planner, and translator for the landlord email I kept postponing. What follows is the full playbook: the myths that waste your money, the room-by-room walkthrough, and the exact prompts to copy. Everything was tested on a real 1970s apartment in Germany, but the method works whether you’re in a Paris studio, a Dutch row house, or a Manchester terrace.
Myth 1: “AI can’t help with anything physical”
This is the myth I believed too. A chatbot can’t feel the draught under your door. True. But here’s what it can do: turn you into a competent draught detective in ten minutes. I photographed my windows and the gap under my front door, described them to Claude (“1970s aluminium-frame windows, rubber seals crumbling, gap under door about 1cm”), and asked what to buy and in what order.
It gave me a prioritised shopping list with the reasoning attached: door draught excluder first (biggest gap, cheapest fix, under €10), then self-adhesive foam window seals, then — only if the first two weren’t enough — secondary glazing film. It also told me what NOT to buy: the magnetic window insulation kits that cost €40 and fall off by December. When I checked its claims against two DIY forums, it was right on both counts. The physical work took me 90 minutes on a Saturday. The AI part took ten.
Myth 2: “The smart thermostat is the big win”
Every energy blog wants to sell you a €200 smart thermostat. Here’s my stance, and I’ll argue it: for most European renters, the smart thermostat is the third thing to buy, not the first. The highest-ROI winter move is sealing air leaks, and AI is the cheapest energy auditor you’ll ever hire.
The numbers: draught-proofing a typical flat costs €20–40 in materials and can cut heating demand by 10–15%, according to multiple European energy agencies. A smart thermostat saves a similar percentage but costs five times more — and in a rented flat, you may not even be allowed to install one. I asked ChatGPT to rank winter upgrades by payback time for a renter in an old building. Its answer: draught-proofing (weeks), radiator reflector foil behind radiators on outside walls (one winter), thick curtains (one winter), smart thermostat (two to three winters). That ranking reshaped my whole plan, and my wallet thanked me.
Myth 3: “You need an expert to read your heating bill”
Nobody teaches you to read a European heating bill, and landlords count on it. I pasted my last annual statement into Gemini — redacted the personal bits first, always — and asked: “Explain this like I’m new to Germany. What’s fixed, what’s consumption-based, and does anything look off?”
It spotted something I’d missed for two years: my building’s heating was billed 70% by consumption and 30% by floor area, which is the legal maximum area-based share in Germany — meaning my careful radiator discipline was only moving 70% of the bill. It also flagged that my “warm water” line was suspiciously high relative to my household size and suggested I ask the Hausverwaltung for the meter readings. One email later — drafted by the AI, in appropriately formal German — I had the readings, and they were fine. But I’d never have known what to ask without the explanation. For the budget side of winter, our guide to building a personal budget with AI pairs well with this.
The room-by-room walkthrough
Here’s the method, tested across two weekends. Do it in this order — each step builds on the last.
The hallway and front door. Walk the flat on a windy day with a lit candle or a strip of tissue paper. Hold it near door frames, window edges, and the letterbox. Where it flutters, air is leaking. Photograph each leak, describe it to your AI of choice, and ask for the specific product type (not brand — types) to fix it. My total: one door excluder, two rolls of foam seal, €23. The hallway stopped feeling like a wind tunnel the same evening.
The living room. This is where you live, so it’s where comfort matters most. Two AI jobs here. First: “My living room is 22m², north-facing, with a radiator under the window. The room never gets above 19°C. What should I check before calling anyone?” The answer correctly identified the likely culprits in order: air trapped in the radiator (bleed it — the little key costs €3), the thermostat valve set wrong, furniture blocking airflow. Bleeding my radiator took four minutes and fixed the room. Four minutes, after two winters of complaining.
Second: ask for a heating schedule. “I work from home three days a week and I’m out 8–18 the other two. Give me a radiator schedule for a north-facing room.” The schedule it produced — lower overnight (16°C is fine for sleeping and saves real money), ramp up an hour before you wake, setback during work-from-home lunch — is the kind of thing an energy consultant charges €150 for.
The bedroom. Bedrooms are where people overheat their homes for no reason. The healthy sleep range is 16–19°C, and every degree above that costs roughly 6% more heating energy. Ask the AI: “I sleep hot/cold, my bedroom faces east, give me a bedding and heating plan.” It’ll do in thirty seconds what took me three winters of trial and error: warm duvet, cool room, window cracked for the last ten minutes before bed. The “cool room, warm bed” principle is the single most underrated comfort trick in European housing.
The bathroom. Mould is the silent winter tax. Warm moist air hits cold walls, and by February you have black spots. The AI’s ventilation plan was specific and — I checked with a building engineer friend — correct: after every shower, open the window fully for 5–10 minutes (shock ventilation, or Stoßlüften as the Germans call it) rather than leaving it tilted all day, which just cools the walls and wastes heat. If you have no window, run the extractor for 20 minutes after showering. This one tip is worth more than any gadget.
The Saturday-morning checklist
Print this. Do it once. Thank yourself in February.
Bleed every radiator. Four minutes each. The key costs €3.
Seal the front door gap. Biggest leak in most flats.
Foam seals on the worst two windows. Not all of them — the worst two.
Shock-ventilate, don’t tilt. Five minutes, wide open, twice a day.
Photograph your meter readings now. Disputes in March need October evidence.
Draft the landlord email about anything broken — today, not in January when everyone’s boiler fails at once.
The landlord email (the one everyone postpones)
Here’s the prompt I used, and the result genuinely impressed me: “Write a polite but firm email in German to my landlord. The living-room window seal is broken and there’s visible condensation between the panes. Request a repair before winter, reference that window maintenance is the landlord’s responsibility, keep it under 150 words.”
What came back was better than what I’d have written: correct formal address, the right legal framing without sounding threatening, and a specific ask with a deadline (“before the heating season begins”). I changed three words and sent it. The repair was scheduled within a week. If you’ve ever stared at a blank email for twenty minutes, this alone justifies the whole exercise. (For bigger home projects, our AI home renovation planner covers the full method.)
What I’d do differently
Honest post-mortem: I should have photographed the meter readings in September, not October — one month of baseline data would have let me measure the draught-proofing savings precisely. And I asked the AI for product recommendations by brand name, which was a mistake; brands vary by country and stock, and the type-level advice (“self-adhesive EPDM foam seal, 9mm”) was far more useful. Ask for types and specs, not brands.
One caution that matters: never paste your full unredacted heating bill or contract into any AI tool. Redact your name, address, customer number, and meter number first. The AI doesn’t need them to explain the bill, and your data shouldn’t be training fodder. (Our European AI privacy checklist covers exactly these habits.)
Two weekends, about €30 in materials, and one flat that finally holds its heat. The thermostat and I are on speaking terms again.
Written by the PromptSkill Team — we test the methods we write about in real European homes, so you get playbooks that survived contact with an actual October, not just theory.