/ InfraIQ

One system for a programme that currently runs on spreadsheets.

InfraIQ is a platform we built in-house to show how monitoring for development-financed infrastructure could work when the whole programme sits in one system instead of a filing cabinet of workbooks and email threads.

/ The problem

The reporting burden is the project risk.

A development-financed infrastructure programme carries obligations that have nothing to do with laying road: environmental and social safeguards, grievance redress, procurement scrutiny, disbursement reporting, and periodic implementation status reporting to the financier. In most programmes each of those lives in its own workbook, maintained by a different person, reconciled by hand before every mission.

Nobody sees the whole picture

Physical progress, spend, and safeguard compliance are tracked separately, so the relationship between them only surfaces when something has already gone wrong.

Reporting is archaeology

Each reporting cycle is reconstructed from scratch out of files of unknown vintage, rather than read off a system that was already current.

Evidence is not traceable

A figure in a report cannot be followed back to the record it came from, which is precisely what an audit asks for.

/ The shape of it

Ten modules. One record underneath.

Every module reads and writes the same programme data model, so a works package, its contract, its disbursement, its safeguard findings, and its position on the map are the same record seen from different angles — not five copies drifting apart.

InfraIQ module map Ten modules — works progress, procurement, financial management, environment, social and grievance redress, disaster response, implementation status reporting, geospatial views, the programme dashboard, and administration — all reading and writing a single shared programme data model. Works progress packages · milestones Procurement stages · red flags Financial disbursement · variance Environment safeguards · findings Social & GRM grievances · redress Shared programme data model ONE RECORD · MANY VIEWS · EVERY FIGURE TRACEABLE TO ITS SOURCE Dashboard programme status Status reporting generated, not rebuilt Geospatial assets on the map Disaster response events · exposure Administration roles · access
Fig. 01 — Modules over a single programme data model

/ What it does

Built around how programmes are actually supervised.

/01

Works progress

Physical progress tracked per contract package against milestones, so schedule slippage is visible while it is still recoverable rather than at the next supervision mission.

/02

Environmental & social safeguards

Safeguard findings and their remediation carried as tracked items with owners and due dates, alongside a grievance redress register that records each complaint through to closure.

/03

Procurement oversight

Packages followed through their procurement stages, with rule-based flags for the patterns supervisors are required to look for.

/04

Financial management

Disbursement against commitment and physical progress in the same view, because spend that runs ahead of works is the signal worth catching early.

/05

Implementation status reporting

Periodic reporting generated from the live record rather than reassembled by hand, with every figure traceable back to the entry it came from.

/06

Geospatial view

Assets, districts, and alignments on a map, so a programme spread across a region can be read spatially instead of as a list of package codes.

/ Intelligence layer

The data is already there. The reading of it is the work.

A programme management system that only stores what people type into it has moved the filing cabinet, not the problem. Once every module writes to one data model, the interesting question is what can be noticed automatically — and what still has to be judged by a person.

How the intelligence layer sits between records and outputs Programme records feed a model layer that detects anomalies, extracts data from submitted documents, classifies grievances and drafts narrative. Every output passes a human review gate before it reaches a report, a flag or a decision. Programme record progress · spend · findings Intelligence layer anomaly detection document extraction classification · routing narrative drafting GATE Human review accept · correct · reject Output report · flag · alert CORRECTIONS RETURN AS TRAINING SIGNAL
Fig. 02 — Nothing reaches a report without passing a person

Notice, don't decide

The model's job is to raise the thing worth looking at — spend running ahead of physical progress, a grievance category spiking in one district, a package stalled between procurement stages. The judgement stays with the supervisor.

Read the paperwork

Contractor submissions, inspection notes and invoices arrive as documents. Extraction turns them into structured records against the right package, so data entry stops being the bottleneck that makes reporting late.

Cite or stay silent

Drafted narrative carries a reference back to the records it came from. A sentence that cannot point at its evidence does not go into a report — which is the difference between assistance and a liability.

/ Reporting

The report should be a read of the system, not a project of its own.

Institutionally funded programmes are held to a reporting rhythm set by the financier, and in most of them each cycle is an event: a fortnight of chasing spreadsheets, reconciling versions, and rebuilding figures whose provenance nobody can quite reconstruct. That work produces no road, no school, and no clinic.

/01

Assembled, not reconstructed

Periodic reporting is generated from the live record. The reporting cycle becomes a review of something already true rather than an exercise in rebuilding it.

/02

Every figure traceable

Each number in a report can be followed back to the entries that produced it, with the timestamp and the person who recorded them. This is what an audit asks for, and it is far cheaper to have by construction than to assemble afterwards.

/03

Written for the reader

A financier, a ministry, and a district office need different cuts of the same truth. Formats differ; the underlying record does not, so the versions cannot quietly disagree.

/04

The same shape elsewhere

Nothing in this is specific to roads. Any programme accountable to someone who funded it — grant-financed health and education, corporate social responsibility portfolios, multi-site operations under a regulator — carries the same problem: distributed activity, central accountability, and reporting that has to be defensible. The architecture transfers; only the vocabulary changes.

/ Why we built it

A capability is easier to assess than a claim.

We could describe our institutional delivery capability in a paragraph. We would rather show a working system, because the questions that decide these engagements — how safeguards are tracked, how a grievance is closed out, how a reported figure is substantiated — are answered by architecture, not adjectives.

InfraIQ is funded and owned by Intelliots. It has not been deployed on a live programme, and nothing on this page should be read as a delivered engagement. If it is close to a problem you are carrying, we will walk you through it under NDA and tell you plainly where the gaps are.

Ask us to walk you through it.

Start a conversation