CostIQ: from a 14,500-line bill of quantities to a live earned-value engine in ten days

Client and context

A Tier 1 mining client is delivering a utility-scale solar photovoltaic construction package as part of its decarbonisation programme. The construction contract runs to roughly 14,500 bill-of-quantities line items across eight control accounts, with priced works in excess of USD 11 million. (Client and project identifiers withheld; figures rounded. Faolan Consulting supports the owner's team on project controls.)

The problem

Cost control on mining construction packages still runs on manual effort. The bill of quantities arrives as a large spreadsheet. Cost engineers re-key it into a control structure, reconcile totals by hand, and only then can an earned-value cycle begin. On a package of this size that is days of skilled effort before the first useful number appears, and every manual touch is a chance to introduce an error that surfaces months later in a disputed payment certificate.

For a specialist consultancy, that manual load is also the ceiling on how many packages one team can control at once.

The approach

FaolanIQ is Faolan Consulting's fleet of AI agents for project controls. The Build pillar, orchestrated by the agent Thor, tasked its cost-intelligence agent Mjolnir with building CostIQ, a cost-control engine, to an approved specification.

The build ran as governed agent-to-agent work: the specification was approved on 2026-06-07, every hand-off between agents travelled as a schema-validated brief with an audit log, and the finished engine had to survive an adversarial review by four independent verifier agents before acceptance. The proven engine shipped on 2026-06-17, ten days after specification approval.

CostIQ covers the core cost-control cycle: a control-account spine, planned and earned value against actual cost, cost and schedule performance indices (CPI and SPI), estimate and variance at completion (EAC and VAC), rules-of-credit progress measurement, payment-certificate assembly, and trend analysis. A shared ingestion front door reads a native bill-of-quantities workbook, validates it, and lands every line in the control structure with its provenance intact. The engine carries 61 automated tests, all passing.

The outcome

• The client's full bill of quantities was ingested through the front door in a single run: roughly 14,500 lines classified into eight control accounts and footed exactly against the priced contract total, with zero manual re-keying.
• Ten days from approved specification to a proven, tested engine; first real client data ingested the following day.
• On the engine's proof fixture, a full earned-value cycle (planned value through variance at completion, for example CPI 0.94 and SPI 0.92 on the fixture control account) runs in seconds. These proof-fixture metrics are illustrative, not client results.
• The first live pilot cycle on the client package was commissioned on 2026-07-06, with baseline results due 2026-07-10.
• The same ingestion front door was reused unchanged on a second, unpriced bill of quantities of roughly 6,500 lines from a different mining megaproject, routing it to FaolanIQ's estimating agent instead. One capability, two products.

What this demonstrates

FaolanIQ agents do not just draft documents; they build, test, and operate project-controls software under governance, with every decision auditable. Database-first delivery means client data lands once, reconciles exactly, and every downstream number traces back to a source line. The result is earned-value discipline at a speed and cost that manual cost engineering cannot match, which is precisely the capacity multiplier a specialist controls consultancy needs.

Controls. Intelligence. Delivered.