The scale exposed the gaps
Saudi Arabia's giga-project portfolio is the largest concentrated construction programme in market. NEOM, the Line, Oxagon, Trojena, the broader Vision 2030 portfolio of projects across the Kingdom, the UAE pipeline of urban-development programmes, the Qatar post-World-Cup development sequence — the scale is unlike anything the global construction industry has handled simultaneously. Oracle Primavera P6 is the institutional default scheduling product across the portfolio. The contractors run it. The owners run it. The programme-management consultancies run it.
The product is mature. The integrator ecosystem is dense. The talent base is well-trained. None of that is the argument. The argument is that the scale of the work is exposing the structural gaps in a product that was designed for a different generation of project complexity. The number of trades, the number of contractors, the number of concurrent fronts of work, the speed at which the design is iterating, and the volume of dependency relationships are operating at a scale the P6 schedule was not built to model in real time.
The market is responding. The Primavera Cloud product is the vendor's modernisation attempt. The Construction and Engineering Intelligence module is the AI surface. The new entrants — Microsoft Project, Asta Powerproject, Phoenix Project Manager, the AI-native schedule-intelligence start-ups — are all positioning against the gaps. The owners and the major contractors are watching the alternatives carefully. The conversation worth having is which workflows the institution should own outright, regardless of which product handles the master schedule.
What Primavera does and where the gaps sit
P6 handles the master schedule. The work breakdown structure, the activity-and-task definition, the resource-loading, the critical-path analysis, the float calculation, the baseline comparison, the earned-value reporting. The product is competent at the schedule arithmetic. The integration to the construction-management tooling — the document control, the design coordination, the cost management — runs through Oracle's broader Construction and Engineering catalogue and through the integrators that have built the cross-platform connectors over the last decade.
The gaps sit in four workflows the product structurally does not perform.
Schedule synthesis at the portfolio level. A giga-project carries 30, 50, sometimes 100 parallel schedules at the contractor level. The owner's programme-management team needs a synthesis across the schedules — the critical-path roll-up, the cross-contractor dependency conflict identification, the resource-contention surfacing at the portfolio horizon. P6 produces this through manual integrator-led roll-ups on a weekly cadence. The owner's question — what is the actual project posture today, against the original baseline, with the cross-contractor dependencies modelled honestly — is answered in a working session, not by the product.
Design-to-schedule alignment. The design is iterating during construction. A change to a particular building, a particular system, a particular trade-front affects the schedule across multiple contractors. The current product surface requires manual update of the affected activities by the contractor's planner, validation by the owner's programme team, and reconciliation against the design package. The lead time from design change to schedule reflection is measured in days. The decisions the operations leadership are making during those days are made against a stale schedule.
Risk and delay-cause attribution. When a delay materialises, the institutional question is what caused it. The contractor's perspective, the owner's perspective, the supervisory perspective, and the dispute-resolution path all require structured delay-cause attribution. The product surfaces the dates. The attribution is human work performed against the schedule history, the change-order log, the design-revision log, and the contractor's correspondence. The work is slow and the conclusions are often contested.
Look-ahead and resource-balancing intelligence. The 90-day look-ahead, the resource-loading conflict identification, the manpower curve smoothing, the equipment-utilisation optimisation — each of these is a workflow the programme team performs in working sessions outside the product, with the product as the data source.
What the agent layer absorbs
The owned agent layer absorbs the four workflow gaps. The architecture is the same pattern we run at every regulated estate.
A data layer that reads the schedules from the contractors through the standard XER and XML exchange formats P6 supports, joins them against the design-package version, the change-order log, the contractor-correspondence catalogue, and the construction-management tooling. The data lives on infrastructure the owner operates.
A schedule-synthesis surface that produces the portfolio roll-up against the current state of the contractor schedules, identifies cross-contractor dependencies that are at risk, surfaces resource-contention conflicts at the portfolio level, and produces the owner's working view in language the programme leadership can act on.
A design-to-schedule alignment agent that watches the design-package version against the schedule reflection, identifies the activities and trades that the design change affects, prepares the structured change-impact assessment, and routes it to the contractor's planner and the owner's programme team. The change-impact reflection runs at the cadence the design iterates at, not at the cadence the integrator-led roll-up runs at.
A delay-cause attribution agent that watches the schedule slip against the documented cause base. When an activity slips, the agent prepares a structured attribution against the change-order log, the design-revision log, the contractor-correspondence catalogue, and the historical pattern of similar slips. The owner's claims team, the contractor's claims team, and the dispute-resolution path all consume the same structured evidence base. The attribution is reproducible.
A look-ahead and resource-balancing agent that absorbs the working sessions the programme team runs outside the product. The 90-day look-ahead is the agent's primary surface. The conflicts and recommendations are the surface the programme leadership reviews and acts on.
What stays with Primavera
The master schedule. The work-breakdown structure. The contractor's working surface for the contractor's planner. The earned-value reporting that the owner's commercial team consumes for the contract administration. The supervisory-level scheduling artefacts that the owner's authority reviews. P6 continues to operate as the institutional schedule of record.
The agent layer reads from P6. The agent layer writes to its own operational surface. The agent layer does not modify the P6 schedule without the programme team's authorisation through the existing change-control process.
The regional dimension
Three dimensions are specific to a giga-project context in this region.
The scale itself. The number of concurrent contractors, the cross-discipline coordination, and the portfolio horizon are larger than any other construction market is handling simultaneously. The agent's portfolio-roll-up loop is the surface the owner's leadership needs to operate against.
The design-iteration cadence. The giga-projects are operating with the design and construction running in parallel at a velocity the traditional sequence does not support. The agent's design-to-schedule alignment loop runs at the cadence the design iterates at.
The commercial dispute exposure. The contract-administration cost across the portfolio is meaningful. The structured delay-cause attribution the agent produces materially changes the commercial conversation at the dispute-resolution path.
The saasinator perspective
We are not arguing against Primavera. P6 is the institutional default for reasons that are defensible. The argument is that the workflows the giga-project portfolio actually runs are workflows the product was not designed to handle at this scale, and the work the programme team is performing manually is the work the agent layer absorbs.
The owner who runs the first agent successfully has changed how the programme operates. The next agent, the next adjacent surface, the next workflow — each is a smaller decision than the first.
What to bring to the diagnostic
The diagnostic for a giga-project engagement is 15 working days. Bring the P6 deployment scope across the portfolio, the contractor's schedule-exchange cadence, the programme-management working-session inventory, and the supervisory and dispute-resolution posture. The output is the workflow recommendation, the architecture sketch, and the first-quarter scope. Book a diagnostic at /diagnostic.