Public planning data, made genuinely usable.

17,112 people are in emergency accommodation in Ireland. Average rent in Dublin is around €2,200 a month. The country needs roughly 33,000 new homes a year and is consistently delivering well below that. Every one of those homes starts with a planning application.

Planning data is technically public. National records are published openly by government. There are 493,616 records across all 31 local authorities. None of them are indexed by any major search engine. None of them are in any AI model's training data. Every council runs its own portal that only lets you look up one record at a time, by reference number you already know.

So planning consultants and architects — the professionals whose job is to make the case for new homes — research precedent by opening one council portal at a time, in tabs, by hand. In the middle of a national housing crisis, the people who build homes can't access the data they need to build them well.

We built it because that gap is closeable with software, today, by anyone who is willing to do the integration work. Nobody had been paid to do it. So we did.

Blindspot Labs started as a hackathon prototype. Three of us spent two weeks making Dublin City Council's planning records searchable, wiring up an AI search interface, and shipping the result.

We won 2nd place at the Nomad AI Competition 2025 in the “Strange Data Project” category, with a €1,500 prize. That isn't the part that mattered. The part that mattered was that planning professionals at the demo day kept asking when it would cover the rest of the country.

So we kept going. The prototype that handled 26,000 Dublin records is now a national platform handling 493,616 records across all 31 Irish local authorities, with a parsed and searchable Dublin City Development Plan, and a hub where users can save searches, save precedents, and group them into projects. What started as a one-council demo became the only product that can answer questions about Irish planning across the entire country.

We are explicit about what software is good at and what it isn't. Software is good at retrieval, indexing, and surfacing patterns across large corpora. Software is bad at applying professional judgement, weighing political context, and being accountable for a decision.

So we focus on the first set. The product gives you faster access to public planning records, faster reference to the development plan, and faster discovery of relevant precedent. Everything that requires judgement — what to argue, how to argue it, whether a particular precedent is genuinely comparable, what conditions to expect, when to escalate to An Bord Pleanála — stays with the professional. Where the AI summarises retrieved material, it cites the reference numbers and policy chapters it drew from, so the professional can verify and quote with confidence.

• Make planning decisions or predict whether a given application will be granted.
• Submit applications to councils on behalf of users.
• Replace qualified planning consultants, architects, or solicitors.
• Anonymise or rewrite public planning records — they are public for a reason and we publish them as the council publishes them.
• Provide legal advice. The platform is a research tool.

Three MSc Computer Science students at Trinity College Dublin. Between us we have prior experience at Microsoft, Intel, IBM, and Workday.

Advised by Kevin Collins of Echofold.ai, who has steered our thinking on what the platform should and shouldn't be.