Forensic schedule analysis →
When the reporting routine has already failed and the question is what the records will support.
Project controls and AI automation is an advisory and delivery service that establishes reliable reporting routines before applying automation to repetitive, rule-bound work. It is for owners, contractors and PMOs that need decision-ready information, with every consequential output traceable to approved project records and reviewed by a named person.
The problem
Most capital projects are not short of data. They have a programme, cost reports, a risk register, progress records and a document register. What is missing is a dependable path from those inputs to a decision. Information is re-keyed, updates are chased by email, and by the time a report reaches a decision-maker the position it describes has moved.That is not simply an efficiency issue. HKA's CRUX Insight research across more than 2,200 projects found sums in dispute averaged 33.4% of contract budgets. A weak routine produces a weak contemporaneous record. The work therefore starts by establishing a control environment people trust, then removes only the repetitive work around it.
Scope
Control framework and reporting routine review
Map what is captured, by whom, in which system, on what cadence, and where the real hand-offs break. The output describes the operating model that exists, not the one in a procedure manual.
Baseline and data standards remediation
Align coding, WBS and cost breakdown structures, calendars, update discipline and the definitions that have drifted between teams.
Document intelligence over approved records
Apply retrieval and extraction to correspondence, minutes, submittals, site instructions and notices. Each result is linked back to its source record.
Automation workflow design
Select high-volume, rule-bound work such as report assembly, update chasing and controlled register checks; exclude contractual judgement from automation.
Capability transfer
Provide runbooks, training and ownership so the operating team can run, adjust and extend the routine after handover.
Method
- 01
Establish what exists
Review systems, data, cadence, ownership and the actual reporting routine.
- 02
Identify the decision that is failing
Automation targets a decision and its evidence, not a fashionable task.
- 03
Stabilise the control environment
Set data standards, update rules, responsibilities and escalation points.
- 04
Build narrow, prove, then widen
Test one workflow against the manual result before extending it.
- 05
Instrument for review
Require source citation, audit logging, a human sign-off point and a defined failure mode.
- 06
Transfer
Document the routine and train the team that will own it.
What you get
Controls and reporting assessment
Current-state map, gap list and prioritised remediation sequence.
Data standards specification
Coding, WBS/CBS alignment, update rules and definitions.
Reporting routine design
Cadence, owners, inputs, outputs and escalation triggers.
Automation candidate register
Candidates assessed for volume, rule-boundedness, judgement content and risk.
Governance and handover pack
Source-citation model, review gates, logging, runbooks and training.
What we need
- Current programme and recent updates
- Current monthly report and underlying source files
- Risk register and document register or DMS export
- Cost report structure and organisation chart for project controls
- Existing dashboards, macros or automations, including abandoned ones
- Data-handling constraints, confidentiality and residency requirements
Engagement
Engagement length and composition depend on the question, the available records and the required decision date. An initial discussion establishes scope; no published prices or generic duration promises are used.
| Engagement shape | Suitable where | Typical duration |
|---|---|---|
| Controls diagnostic | A PMO that cannot yet locate the reporting failure. | Confirmed after scope and record review |
| Controls remediation and routine design | A project with data in place but no dependable path to a decision. | Confirmed after scope and record review |
| Automation pilot | One workflow designed, tested and handed over end to end. | Confirmed after scope and record review |
| Advisory retainer | Ongoing controls oversight and incremental automation. | Confirmed after scope and record review |
| Capability programme | Training-led transfer to an in-house team. | Confirmed after scope and record review |
Scope limits
- Software licensing, resale or vendor implementation contracts.
- Replacing project governance or contractual decision-making.
- Automating a notice, determination or entitlement assessment without named human review.
- Processing records that the client is not permitted to make available.
Frequently asked questions
No. Automation is applied to repetitive, rule-bound work so the controls team can spend its time on analysis and judgement. Every output with contractual or commercial consequence passes through a named human reviewer before it is relied upon.
Systems are rarely the constraint. The constraint is usually ownership, cadence, data standards and the action taken when an update is late. Automation over an unstable control environment produces faster unreliable reporting.
It depends on the data-handling constraints and, where a dispute is live, legal advice. Permissioned document sets, access controls and audit logging are established at the beginning of the engagement.
The workflow is constrained to an approved record set, requires source citations and surfaces unsupported answers as gaps rather than guesses. Human review remains the control.
Capability transfer is a named workstream. Runbooks, standards and training are provided so the team can operate and extend the process independently.
Initial discussion
Discuss a project controls and ai automation challenge.
Share the project context, the decision required and the records available. The first conversation is used to establish whether the work is a fit.
Automation is introduced as support for accountable project judgement, not as a replacement for project governance.
Last reviewed: August 2026