Project Management Offices traditionally produce and maintain a considerable collection of management artefacts: schedules, progress reports, risk registers, issue logs, action trackers, change registers, dashboards and meeting minutes.
These serve legitimate purposes, but they also introduce an unusual characteristic into project management. Much of the effort goes into producing representations of the work rather than managing the work itself.
A supplier reports progress. That information is entered into a tracker, incorporated into a schedule update, summarised in a progress report and ultimately presented on a management dashboard. The same underlying event may be interpreted, transferred and rewritten several times.
Each artefact becomes another version of reality that needs to be maintained and reconciled.
What if these separate management artefacts were no longer necessary?
The underlying records remain
A project still needs authoritative information: contracts, approved scope, budgets, baseline milestones, change instructions, supplier commitments, progress observations, test evidence, decisions and records of completed work.
These are not reporting artefacts. They represent the actual obligations, events and evidence from which the state of delivery can be established.
AI introduces the possibility of maintaining the relationships between these records continuously.
A contractual change can be associated with its affected scope, cost, schedule activities and approvals. A progress observation can be linked to the relevant work package. A delayed delivery can be connected to dependent activities and forecast milestones.
Instead of maintaining a separate risk register, issue log and monthly progress report, the system could derive these views from the underlying information.
A risk would be a recognised uncertainty associated with specific activities and consequences. An issue would be an observed condition requiring action. A programme report would be a view of the current delivery position, generated from the available evidence.
The distinction is important: the information is maintained, but the documents describing it no longer need to be maintained independently.
From reporting cycles to continuous understanding
Traditional reporting often creates a delay between events occurring and their significance becoming visible.
A continuously maintained delivery model could identify new dependencies, emerging risks and forecast changes as information becomes available.
Programme managers could ask for the current position of a milestone, the reasons behind a forecast movement or the commercial implications of outstanding work.
The system would assemble the answer from the relevant records, including its assumptions, sources and uncertainties.
Formal reports and contractual submissions would still be produced when required. The difference is that they would be generated as dated, auditable snapshots of the underlying information rather than assembled and subsequently maintained as separate sources of truth.
Historical snapshots would remain available, preserving what was known and reported at a particular date.
Capturing progress directly from the work
One particularly interesting opportunity is to reduce the need for people to report progress manually.
The Base-e venture, AI Field Intelligence System Using Video, explores how video captured in operational and construction environments could be interpreted as structured progress information.
For example, suitably positioned cameras could provide timestamped observations of equipment arriving, installation activities taking place or visible physical work reaching particular stages.
AI could classify these observations and associate them with relevant work packages, locations and planned activities.
This would provide another source of delivery evidence, potentially reducing the dependence on manually prepared progress updates.
However, observing installed equipment does not establish that it has been tested, accepted or commissioned. Video captures only what is visible, and its interpretation may be uncertain.
A dependable system would combine video-derived observations with supplier records, inspections, test results and formal acceptance evidence, distinguishing observed progress from verified completion.
The video intelligence project is exploring one component of this wider concept, not an already deployed replacement for project controls.
What becomes of the PMO?
If management information can be generated directly from a shared, continuously maintained record of delivery, much of the administrative production work traditionally undertaken by a PMO could disappear.
But the responsibility for project controls would not.
Someone must still establish baselines, challenge forecasts, approve changes, validate evidence, manage contractual obligations and decide how to respond to emerging problems.
The PMO could become a smaller, more specialised capability concerned with the integrity of information, the reliability of forecasts and the effectiveness of delivery decisions.
The underlying architecture would need robust source records, persistent identifiers, version history, access controls and clear distinctions between reported claims, AI inferences and authorised facts.
Without these controls, automating reporting could simply generate incorrect information more quickly.
The opportunity is to stop maintaining multiple representations of delivery and instead maintain the evidence from which any required representation can be produced.
Project management could move away from the periodic manufacture of reports and registers towards continuous interpretation of what is actually happening.
The documents become outputs. The underlying delivery record becomes the system.