Build-in-Public Studio

Summaries

Per-project product summaries. Two modes: website (tagline, intro, feature list — ready for a landing page) and launch announcement (X thread + Indie Hackers long-form, with the same voice rules as your weekly drafts).

Project

Pick a watched repo to generate or view summaries for it.

build-in-public-studio PanelProBuild Property-Reconfig-App WyCo-Irrigation_System stashery WyCo-Trace-Maps bloomr

Website summary

Tagline

AI-powered property reconfiguration analysis for UK investors

Intro

Property Reconfig App analyses UK property listings and tells you whether a reconfiguration — converting a house into flats, adding a floor, splitting rooms — actually pencils out. You paste in a listing, the app pulls Land Registry data, enriches the address, runs a multi-pass AI analysis, and returns ranked reconfiguration ideas with cost estimates and a BRR (Buy, Refurbish, Refinance) stress test. It's built for property investors who want a faster first-pass filter before spending money on solicitors or surveyors. The analysis streams in live so you're not staring at a spinner, and every idea comes with a confidence score so you know how much weight to put on it. The entire codebase is 100% AI-generated using Claude. No professional developers were harmed in the making of this app.

Features

•Multi-pass AI analysis scores and ranks reconfiguration ideas with per-idea cost estimates and confidence ratings.
•Live-streamed analysis results appear card by card — no waiting for the full run to finish before you see output.
•BRR stress test models cash-in-deal figures and refinance outcomes for Buy, Refurbish, Refinance strategies.
•AI-generated Before / After room renders using FLUX Kontext Pro img2img when a listing photo is available.
•Automatic Land Registry lookup and address enrichment pre-populate every analysis from a single listing URL.
•Annotated floor plan overlay marks reconfiguration opportunities directly on the property's floor plan image.

Launch announcement

1/ I built a property reconfiguration analyser — and I wrote zero of the code. Every line is AI-generated. I'm not a developer. This is what happened when a complete beginner shipped anyway. 2/ The app analyses UK properties for reconfiguration potential: BRR stress tests, floor plan annotation, AI-rendered Before/After room visuals. Things I had no idea how to build 3 months ago. 3/ The AI render piece was the wildest rabbit hole. Getting Claude to classify a room, then FLUX Kontext Pro to do img2img on the "Before" photo, then auto-retry on failure — that's three models chained together. It actually works. [VERIFY: consistency across room types still being tested] 4/ The hardest bug this week: floor plan markers kept escaping their bounding box. The fix was making the model declare a drawing_bounding_box first, then clamping all coordinates server-side. Simple in hindsight. Took two full sessions to find. 5/ The analysis pipeline runs 4 passes. Pass 3 was silently failing when there were too many ideas — it was hitting a token ceiling and returning nothing. Fix: scale max_tokens with idea count, and fail loudly instead of quietly. Loud failures are a gift. 6/ I had a real user review session lined up (hi Vicky). The day before, I parallelised two analysis passes, rewrote auth, added a brand strip, and built a Before/After carousel. Deadline pressure is a cheat code. 7/ Everything you see at [LINK] was written by Claude. I directed it, debugged it (slowly, with a lot of Googling), and shipped it. If you're a non-technical person sitting on a product idea, this is proof the barrier is lower than you think. 8/ Would love feedback — especially from anyone in UK property who'd actually use this. What's missing? What's broken? What would make you come back? [LINK]