How Foreland works, plainly.

Foreland works from public planning information published by Irish local authorities and national public datasets: planning application records — reference, location, proposal description, decision, key dates and appeal outcome where one exists — covering all 31 local authorities.

Alongside the application record, Foreland indexes the text of adopted city and county development plans. Today that means the Dublin City, Cork County and Galway City development plans, with more being added; each excerpt keeps its plan, chapter and page so it can be cited precisely.

Foreland does not use private or paywalled data, and it does not scrape personal information. What you see is a restructured view of what the public record already says.

When you ask a question, Foreland searches the indexed record by meaning, not just by keyword — so an application described as a “two-storey rear extension” is found even if your question says “ground and first-floor addition”. Each retrieved record carries a relevance score, and when the record is a weak match for the question, the answer says so rather than presenting thin evidence with false confidence.

Questions about aggregates — approval rates, decision times, trends per authority — are answered from statistics computed over the full dataset, not from the handful of retrieved records, so a “what share of applications are refused” answer reflects the whole record.

The browsable precedent database and map cover the full dataset. AI answers are weighted towards recent decisions, because recent decisions are the most reliable guide to how an authority decides today. See limitations below.

Answers are generated by AI that is given the retrieved planning records, relevant development-plan excerpts and the dataset statistics — and is required to answer only from that material, to cite the specific records it draws on, and to say when the material doesn't support an answer.

The AI's job is synthesis and explanation; the planning record's job is to be the source of truth. This reduces — but does not eliminate — the risk of an answer stating something the sources don't support, which is why every answer carries its citations and why we ask you to review them.

Planning records are cited by their register reference (for example, Ref: 3141/21) and each cited record is listed under the answer with its location, decision and a relevance score. Opening a citation shows the underlying register entry inside Foreland.

Development-plan material is cited by plan, chapter and page (for example, Dublin City Development Plan, Ch. 5, p. 142) with the quoted excerpt shown, so you can check the policy text in context.

Uncertainty is treated as a finding. When retrieval confidence is low, when comparable applications are still undecided, or when the record around a site is thin, that is stated in the answer rather than smoothed over.

Foreland distinguishes three things visually and in language: facts from source records, Foreland's synthesis of those facts, and gaps where evidence is missing or pending. AI-generated synthesis is labelled as such.

Before a finding informs a professional decision: open the cited records and read the decision text; check the reference against the local authority's own register or An Bord Pleanála for the authoritative, current version; and read the full planner's or inspector's report where a refusal or appeal matters to your case.

Foreland supports planning research and early-stage decision-making. Records and AI-generated analysis should be checked against official sources before professional reliance — Foreland is not legal or planning advice.

Coverage and freshness have real limits, and it is better you know them than discover them:

• The dataset is refreshed periodically from the national feed, not in real time. A decision issued yesterday may not be reflected yet; the record count on the homepage reflects the current dataset.
• AI answers are weighted towards recent applications. Older records remain searchable and browsable in the precedent database, but may not be surfaced in AI answers.
• Digitisation quality varies by authority and era — records before the mid-2000s are inconsistent, and proposal descriptions and refusal summaries differ in detail from council to council.
• Development-plan search currently covers three adopted plans (Dublin City, Cork County, Galway City); questions touching other authorities' policies will not have policy citations yet.
• Geocoding covers most but not all records, so map views can omit records whose addresses could not be located.