493,616 Records: What National Planning Coverage Reveals
We pulled every planning record from every Irish local authority and put them all in one searchable index. Here's what showed up that nobody had been able to see before.
By Faith Olopade
We finished a job last week that had been sitting on every planning department's desk for years: get every planning application from every local authority in Ireland into one searchable place.
The number, when the scrape finished, was 493,616.
That's the count across all 31 local authorities, going back to 2003. Cork County alone has 126,300 records. Dún Laoghaire-Rathdown has 37,864. South Dublin 32,793. Dublin City 26,422. The smallest is Leitrim with 2,013. Some of these you'd guess. Others surprised us.
This post is about what we saw when we put them all together for the first time.
The data was always there
None of this was scraped from anywhere private. Every record on Blindspot Labs is a public planning application, published under the Planning and Development Act 2000, available through official government publication. You don't need a key. You don't need to be a journalist. You don't need an FOI request.
You need software, patience, and the willingness to reconcile 31 different naming conventions for what is fundamentally the same field.
That's the entire trick. The data was open. It just wasn't usable.
What "all 31 councils" actually changes
When you can only see one council, certain questions are unanswerable. Where in the country are extension applications most often refused? Which councils are fastest? Which approve the most large-scale residential? Which authorities never see a single demolition application?
Until last week we didn't know either, because nobody had put them next to each other in a single index.
A few patterns showed up immediately.
Big counties dwarf big cities. Cork County (126k) has more records than Dublin City (26k), Dún Laoghaire-Rathdown (38k), and South Dublin (33k) combined. Not because Cork is busier — because county boundaries cover hundreds of villages and rural permissions, where city authorities cover a small dense footprint and most extensions don't trigger a planning application.
Some councils have under 5,000 records total over 23 years. Sligo, Carlow, Galway City, Longford, Leitrim. If you're a planning consultant working in those areas, your local-precedent corpus is genuinely small. Cross-council comparison stops being a luxury and starts being the only way to do precedent research at all.
The big-data council is also the most opaque. Cork County's 126k records is roughly 5x what Dublin City has, but Cork County's planning portal is harder to search and does not return more than one record at a time. We've made that data 100x more accessible just by putting it in one searchable place. We have not changed a single byte of it.
"Searchable" is doing a lot of work in that sentence
The records aren't just in a database. The search understands meaning, not only keywords — "rear extension on a protected structure" finds the right records without you having to know the exact planning vocabulary a given council uses. The full corpus backs the table and map views; AI answers draw on the record with a deliberate weighting towards recent decisions, because recent decisions are the best guide to how an authority decides today.
The point is that AI can finally see Ireland's planning data at all. ChatGPT, Claude, Gemini — none of these models have any of this in their training data, because no search engine ever crawled it. The publishing format is machine-facing and query-driven; crawlers follow links, and there were no links to follow.
So before this week, if you asked any AI model "what extensions were refused in Rathmines in 2024?" you got a polite refusal or a hallucination. Now you get five real reference numbers from real Dublin City Council files, with the actual decisions on each.
The bigger pattern
Planning is the most visible example of a much larger problem. Every layer of Irish public administration produces public data that is technically published but practically invisible: licensing data, building control records, procurement contracts, environmental impact statements, derelict-sites registers, An Bord Pleanála inspector reports. Each one sits behind a different fragmented authority, in a different format, behind a different paginated API or PDF archive.
It's all open. It's all useless until someone bothers to integrate it.
We started with planning because the housing crisis is the place where the cost of inaccessible data is most directly paid. But the recipe is the same for every other dataset on this list. Get the records. Put them in one place. Make the format usable. Index them so an AI can read them.
That's what we did. 493,616 records is just the first one.