Skip to main content
lab-notes.aiPublic-service AI playbooks
Opportunity 13/17TransportTransport

Road Maintenance Planning

Explore how road defects move through inspection, prioritisation, and repair, and what published records cannot settle.

Inspect starter dataSet up the build partner
Published sources
2
Starter data
20 synthetic records
Domain build partner
build-road-maintenance
Last reviewed

From the draft strategy

Opportunity

The draft strategy names AI for road management as a potential public-service application, describing early detection of defects and prioritising repairs, without saying what imagery it means, which inspection standard applies, or how money would follow the ranking.

Where the draft says it

Table 2 — potential public-service applications

Read the draft strategy source (opens in a new tab)

Source register

Research already done

Real published source

We reviewed these published sources so you do not have to start from zero. Each source shows what it covers, how it can be accessed, and why it matters.

2 sources

  1. Department for Infrastructure

    Open

    Northern Ireland road network and condition statistics: technical report

    What it covers
    How the department measures its network and its condition: road length by class — motorway, A, B, C, and unclassified — machine condition surveys that produce a road condition index, and surface defects recorded during routine safety inspections in the Road Maintenance Client System, which also ranks the repairs.
    Why it matters here
    It is the source of this playbook's road classes and defect vocabulary, and it states the limit that matters most: a defect only exists in the figures once an inspection has recorded it.

    Open the source (opens in a new tab)

  2. Department for Infrastructure

    Open

    Potholes recorded by DfI Roads on the public road network each year since 2020

    What it covers
    A published release of pothole counts on the Northern Ireland public road network by year, separating those reported by the public from those found by inspection, and those repaired from those still waiting.
    Why it matters here
    It is the published shape of the queue this playbook's synthetic records imitate, which is why each record carries a status rather than a score.

    Open the source (opens in a new tab)

Starter data

Starter dataset

Synthetic working data

Use 20 synthetic defect records to explore triage without using a road authority's operational records.

AI-assisted research helped identify and interpret the published sources. We then created a small, non-sensitive synthetic dataset shaped by the information those sources expose. It is for exploration—not evidence, training, or operational decisions.

Records
20
Fields
6

Field preview

table view
  • id

    SYN-RM-01 · SYN-RM-02 · SYN-RM-03

  • roadClass

    Motorway · A · B

  • defectType

    Rutting · Cracking · Pothole

  • severity

    R2 · R3 · R1

  • reportedWeek

    2025-W09 · 2025-W11 · 2025-W10

  • status

    repair instructed · inspected · repaired

Inspect all 20 recordsView repository file (opens in a new tab)Download JSON (opens in a new tab)

How it was prepared

AI authored fictional defect types, report weeks, and queue statuses using published road classes and R1 to R3 severity codes.

Limitations

  • The records are invented, so nothing here describes a defect, a road, or a repair backlog anywhere in Northern Ireland.
  • There is no imagery behind any of these rows, so this file cannot say anything about whether a classifier would recognise a defect on a real surface, in real weather, from a real camera.
  • A severity code and a status leave out cost, repair method, and what else on the network is competing for the same crew.

Clone and build

Domain build partner

Build with a domain-aware coding agent

Clone the repository, then ask your coding agent to load the skill that ships with it. The skill brings the opportunity, sources, starter data, known unknowns and constraints into the conversation before anything is proposed. It follows the open Agent Skills standard, so it works in Claude Code, Codex and any other agent that reads skills.

01 / Clone

git clone https://github.com/Hypership-Software/lab-notes.ai.git

02 / Invoke

Claude Code
/build-road-maintenance
Codex
$build-road-maintenance

Conditions, not a checklist

Before you build

These are not footnotes. They are conditions that any responsible prototype would need to address.

  • Both missed and false defects carry costs

    A missed defect is a safety problem, and a false one spends inspection time that was already short.

  • Recorded defects reflect inspection coverage

    Defects only enter the figures once someone has inspected or reported them, so places that are surveyed and reported less can look as though they need less.

  • Image classifications do not choose repairs

    Shadows, standing water, road markings, and old repairs all look like damage in an image, and a classification decides nothing about repair method, cost, or what the network needs most.