Start a Project →
03 / Agentic AI

An AI back office that reads everything, forgets nothing

Not chatbots — agentic AI agents that work through the same documents our people do, at a scale no human team can match. Your project directors spend their hours on judgement, not paperwork.

Drawing register reconciliation

Every sheet read by our agents and reconciled against the register — missing sheets, superseded revisions and unlisted issues flagged before tender close.

Full-set review in hours, not the week a manual check takes.

Clarifications & RFI intelligence

Cross-discipline review that builds the clarifications register automatically — clashes, gaps and ambiguities logged with a snapshot of the exact drawing zone.

Issues surfaced pre-contract, where they cost dollars instead of thousands.

Tender & scope cross-referencing

Scopes of works checked line-by-line against clarification registers, specifications and addenda — nothing cited that doesn't exist, nothing agreed that never made it into the scope.

No scope gap rides through to become a variation claim.

Programme risk scanning

Contractor programmes interrogated for hidden logic, soft float and unrealistic durations — with plain-English risk notes your steering committee can act on.

Slippage visible months early, while recovery is still cheap.

Live demonstration

Tender documents to a signed recommendation

The agents read the tender documents, draft the scope of works, then assess what comes back. The recommendation waits for a person.

Everything issued for tender, read end to end — 0 clauses indexed and cross-referenced.

Specification142 pp · 9 trade sectionsQueued
Drawing set14 sheets · Rev C–EQueued
Addenda 01–03issued 2 Jun – 14 JulQueued
Head contractstandard form, amendedQueued
Preliminariessite access, hours, hoistingQueued
Clarification register31 responsesQueued

A scope of works written from the documents themselves. Every line carries the clause or sheet it came from.

01Preliminaries & site establishmentPrelims cl. 2.4 · Sheet A-0101
02Demolition & make goodSpec 02 41 00 · Sheet A-1201
03Partitions, linings & ceilingsSpec 09 21 00 · Sheets A-2401, A-3010
04Mechanical servicesSpec 23 00 00 · Sheets H-4101, H-4102
05Electrical, data & securitySpec 26 00 00 · Sheets E-5101, E-5102
06Fire servicesSpec 21 00 00 · Sheet F-6101
07Hydraulic servicesSpec 22 00 00 · Sheet P-7101
08Finishes, joinery & loose itemsSpec 09 60 00 · Sheet A-9002
09Exclusions & clarificationsClarification register items 4, 11, 27

Nine sections, every line cited. Reviewed by a project director before it goes out with the tender package.

Three submissions, each read against the scope of works and the clarification register rather than against each other's summaries.

Tenderer A$4.82M38 weeks
— / 100

Excludes after-hours hoisting priced by the others. Not a like-for-like price.

Tenderer B$4.96M34 weeks
— / 100
Tenderer C$4.71M41 weeks
— / 100

Programme shows no float to practical completion and omits the base-build shutdown.

Draft recommendation

Award to Tenderer B — $4.96M, 34 weeks

  • 01The only conforming price. A and C both carry exclusions that would return as variations.
  • 02$140k above the lowest, against $310k of scope the lowest has excluded.
  • 03Shortest programme, with float shown to practical completion.
Awaiting project director sign-off

The agents got it to this point in an afternoon. The decision, and the name on it, stayed with a person.

// ILLUSTRATIVE TENDER — PRICES, PROGRAMMES AND FINDINGS ARE EXAMPLES, NOT A REAL PROJECT

STEP 1 · AI AGENTS

Ingest

Drawings, registers, programmes and contracts read by our agents on receipt — every revision, every addendum.

STEP 2 · AI AGENTS

Cross-check

Every document reconciled against every other. Discrepancies don't hide between disciplines.

STEP 3 · AI AGENTS

Flag

Findings logged with evidence — the exact sheet, clause or programme line, snapshot attached.

STEP 4 · HUMAN

Decide

A project director reviews every flag. Judgement stays human; the agents just make sure nothing is missed.

Deep dive: how agentic AI is changing project management →