FlipHus

How the renovation analysis works

Overview

Every listing on FlipHus is automatically analysed by AI to estimate its renovation potential and cost. The system examines up to 6 photos per listing, identifies specific renovation indicators, and produces a score from 1-10 along with a room-by-room cost breakdown.

The goal is simple: help investors instantly distinguish between a dated apartment with high upside and a move-in ready unit — without visiting every property in person.

The process

1

Collect listings

FlipHus monitors Hemnet daily for new apartment listings in target areas under 4M SEK. Each listing's photos, price, size, and room count are collected automatically.

2

AI photo analysis

Up to 6 photos per listing are sent to Claude (Anthropic's AI) for visual analysis. The model is trained to recognise specific Swedish apartment renovation indicators — the same things an experienced renovator would look for on a viewing.

3

Room-by-room assessment

Each visible room is classified by type (bathroom, kitchen, living area) and condition (original, dated, fair, renovated, newly renovated). Specific indicators found are listed — e.g. "linoleum flooring", "white range hood", "dated tile pattern".

4

Cost estimation

Based on the room types, conditions, and the listing's stated size, the AI estimates renovation cost per room and in total. Estimates use current Swedish contractor market rates.

5

Scoring and ranking

Each listing gets a renovation score (1-10), a confidence level, and an investment potential rating (high/medium/low). Listings can be sorted and filtered by these scores.

What the AI looks for

These indicators are specific to Swedish apartments, particularly miljonprogrammet-era housing from the 1960s-1970s.

Bathroom
  • Linoleum or vinyl flooring
  • Blue flooring or blue tiles
  • Dated or cracked tile patterns
  • Old fixtures (basin, toilet, shower)
  • Mould or water damage signs
Kitchen
  • White plastic range hood
  • White non-integrated dishwasher
  • Shiny metallic sink/bench combo
  • Laminate countertops
  • Dated cabinet fronts
  • No integrated appliances
General
  • Worn or dated flooring
  • Old radiators
  • Dated wallpaper or paint
  • Old electrical outlets/switches
  • Original 60s/70s/80s interior

The renovation score

The score reflects how much renovation work the AI detected, from 1 (fully renovated) to 10 (everything needs replacing).

1
2
3
4
5
6
7
8
9
10
ScoreLabelWhat it means
1-3Move-in readyRecently renovated or in good condition. Minimal upside from renovation.
4-6Some potentialMixed condition — some rooms updated, others dated. Selective renovation could add value.
7-10High potentialClearly unrenovated. Original kitchen, bathroom, flooring. Maximum renovation upside.

Cost estimation ranges

These are approximate renovation costs based on current Swedish market rates for standard contractor work. Actual costs vary by scope, finishes, and contractor.

Room / scopeTypical conditionEst. cost (SEK)
Bathroom (full)Original linoleum, old fixtures, dated tiles120 000 - 200 000
Kitchen (full)Old cabinets, laminate tops, white appliances80 000 - 180 000
Flooring (full apt)Worn parquet or vinyl, 50-70 m²40 000 - 80 000
Paint / wallpaperDated throughout15 000 - 35 000
ElectricalOld outlets, switches10 000 - 25 000
Typical full renovation (2-room apt)250 000 - 450 000

Confidence level

Each analysis includes a confidence score from 0 to 1. This reflects how certain the AI is about its assessment:

0.8-1.0 — Clear photos showing key rooms (bathroom, kitchen). High confidence in score and cost estimate.

0.5-0.8 — Some rooms visible but not all, or photos are at awkward angles. Score is directionally correct but costs may be less accurate.

Below 0.5 — Limited or unclear photos. Treat the score as a rough indication only.

What makes a deal work (or not)

A high renovation score alone doesn't mean a good investment. FlipHus calculates estimated profit using these factors:

FactorWhat to look forImpact
Current sqm price vs area averageIf asking price per m² is already near renovated average, there's no upsideDeal-breaker
Apartment sizeSmaller apartments (<45 m²) tend to have higher sqm prices, squeezing marginsHigh
Monthly fee (avgift)High avgift = high carrying cost during renovation. 5K/mo × 7 months = 35K goneHigh
Days on market200+ days = something is wrong (price, BRF issues, location). Also indicates low liquidityMedium
Stambyte statusIf stambyte isn't done, cosmetic renovation won't add enough value. BRF may also raise avgiftDeal-breaker
BRF debt per m²High debt = risk of future avgift increases. >10K/m² is a warning signMedium
Renovation scopeKitchen + bathroom + flooring = highest ROI. Kitchen-only or paint-only = lower upliftPositive

Area liquidity

How quickly properties sell in each area matters for the exit strategy. A renovation that takes 6 months but then sits unsold for another 6 doubles your carrying cost.

Farsta — Strong liquidity. High demand from young professionals and families. Well-connected via Green line metro. Properties typically sell within 30-60 days.

Rissne — Good liquidity for renovated units. Unrenovated apartments may take longer. Blue line metro gives fast access to Sundbyberg centrum and the city.

The days on market badge on each listing helps you gauge this — listings sitting for 100+ days in a normally liquid area suggest pricing or property-specific issues.

Renovated price benchmarks

FlipHus estimates post-renovation sale prices using area-specific sqm averages derived from recently listed renovated apartments (renovation score 1-3) in the same area.

AreaRenovated avg. kr/m²
Hökarängen60,000
Enskede / Gubbängen52,600
Sköndal50,400
Bromma50,000
Farsta45,600
Rissne / Sundbyberg42,500
Kista38,000
Fagersjö31,000

These benchmarks are updated regularly. A listing is flagged as unprofitable when its current asking price per m² is already near or above the renovated benchmark for its area.

Technology

FlipHus uses Claude by Anthropic — a state-of-the-art multimodal AI that can analyse images and text together. The model receives listing photos alongside a detailed prompt calibrated for Swedish apartment renovation assessment.

Each listing analysis examines up to 6 photos in a single request, allowing the AI to build a holistic view of the apartment's condition rather than assessing rooms in isolation.

The system prompt has been specifically tuned for Swedish construction standards, miljonprogrammet-era building characteristics, and current renovation market rates in the Stockholm region.