October 2026
The Future of AI in Project Controls is Not Autonomous. It’s Accelerated.
AI is unlikely to remove the need for schedulers, estimators or project controls professionals anytime soon. What it can do is compress the work around them: turn days of preparation into hours, make large volumes of project data easier to interrogate, and give experienced teams a much better starting point.
That is the opportunity Dr. Burcin Kaplanoglu, Head of Oracle Industry Labs, sees taking shape across construction. In one example, Kaplanoglu describes how AI could read an RFP, identify key project requirements, ask follow-up questions and generate a preliminary construction schedule. A scheduling exercise that typically takes the project team three to five days could potentially be reduced to about three hours.
The important point is not that AI creates the schedule. It is that the scheduler starts three steps ahead.
“Again, humans are still involved. There is human factor. Schedulers are very, very important.”
That distinction runs through nearly every promising AI use case in project controls today. The technology can prepare, organize, search, analyze and suggest. The people who understand the project still have to decide whether the output makes sense.
For capital project teams, that changes the question from “Will AI replace project controls professionals?” to something much more useful: How much more valuable could those professionals become if AI removes the work that keeps them from applying their expertise?
Potentially from Days to Hours
Scheduling provides one of the clearest examples. When a contractor receives an RFP, the project team has to interpret requirements, understand scope, consider similar work, establish assumptions and begin building a schedule. Much of that early work is necessary, but not all of it requires the highest-value judgment of an experienced scheduler.
Kaplanoglu sees AI changing that starting point. An AI system could ingest the RFP, recognize major project elements, ask clarifying questions and generate an initial construction schedule for the scheduler to review. A process that might normally consume several days could potentially happen in several hours.
The scheduler still determines whether the sequencing is realistic, whether assumptions hold and whether the schedule reflects how the project will actually be built. But instead of spending most of that time creating version one, the scheduler can spend it making version one better.
That is a much more practical way to think about AI in project controls. Its value is not simply automation. Its value is moving experts closer to the decisions where their judgment matters most.
The Same Shift is Coming to Estimating and Forecasting
Scheduling is only one example. Estimating follows a similar pattern. Teams often spend significant time recreating work to validate an estimate, compare scenarios or check assumptions. AI can potentially provide an independent starting point or verification layer, giving an estimator something to challenge rather than requiring another person to rebuild the work from scratch.
The expert still owns the answer, but less expert time has to be spent simply creating another version of the same work. The same logic applies to AI-assisted cost forecasting. Project controls teams already have historical costs, schedules, budgets, trends and performance data; the challenge is often getting from that information to an answer quickly enough to matter.
AI can help surface where cost trends are changing, where schedule risk may be emerging or where a forecast deserves a closer look. But the professional still has to understand why the signal matters and what should happen next. That is the pattern worth watching: AI does more of the preparation. Experts do more of the interpretation.
Project Teams May Soon “Talk” to Their Data
One of the more important changes may be much simpler than a fully automated forecast or schedule. Project data is notoriously difficult to access. Cost lives in one place. Schedule in another. Field records somewhere else. Reports are assembled manually, and answers may depend on who knows which spreadsheet, dashboard or system to open.
Large language models create the possibility of a much simpler interface. Kaplanoglu describes a future where project teams can effectively “talk to your cost, talk to your schedule, talk to your plan, talk to your budget.”
In practice, a project leader could ask which activities are showing the greatest schedule risk, where actual cost is beginning to diverge from plan, which projects need management attention this week, what changed in the forecast, or what happened on similar projects at the same stage.
Some of these questions can already be answered today. The difference is how much effort it takes to get the answer. Instead of requiring every user to understand the underlying data structure or reporting interface, project teams may increasingly be able to ask for what they need in plain language. Better access to information can bring more people into the decision-making process without requiring them to become experts in the system itself.
But “Mostly Right” is not Good Enough on a Billion-dollar Project
This is where enthusiasm around AI needs some restraint. Construction and capital projects are not forgiving environments for plausible answers. An AI-generated schedule can look reasonable and still miss a critical sequencing constraint. An estimate can appear complete and still overlook something expensive. A progress calculation can be directionally right and still lead management to the wrong conclusion.
Kaplanoglu points to this challenge in computer vision. Getting a model to a high level of accuracy is one thing. Closing the final gap is much harder, and in engineering and construction, that final gap matters.
A recommendation engine suggesting the wrong movie creates a minor inconvenience. A project controls system producing the wrong answer can influence decisions involving millions or billions of dollars. That is why human review is not simply a temporary limitation while AI gets better. It is part of the operating model.
Experienced project professionals provide something the model does not automatically have: context. They know whether an output passes the smell test. They understand why one project differs from another. They recognize when an assumption that worked previously does not apply here. AI may become very good at generating the first answer. Project controls professionals still have to know whether it is the right one.
Better AI Makes Data Quality More Important, not Less
The quality of the answer also depends on something far less glamorous than AI: the quality of the underlying information. Kaplanoglu saw this firsthand through Oracle Industry Labs’ work with drone-based data capture. Small differences in flight path, camera angle, image quality or whether a target was fully captured could materially affect what the system was able to recognize.
Kaplanoglu’s conclusion is straightforward:
“Quality, quality, quality—data quality really matters.”
The same principle applies across project controls. If cost codes are inconsistent, schedule activities are named differently across projects, historical data lacks context or teams maintain competing versions of the forecast, AI does not make those problems disappear. It inherits them. In some cases, AI may expose weaknesses that teams have been compensating for manually for years.
That makes data readiness one of the most important parts of AI readiness. Before asking what AI can do with project data, leaders need to understand whether that data is complete, consistent and trustworthy enough to support the decisions they want AI to improve.
A useful starting point is to ask:
- What decision or workflow are we trying to improve?
- What information does AI need to support it?
- How complete and consistent is that information today?
- Can an experienced project professional trace and challenge the output?
- What happens when the system is uncertain or wrong?
Those questions tell you much more about whether an AI use case is viable than an impressive demonstration ever will.
AI Can Make Messy Project Information Useful, Too
That does not mean every useful AI application requires perfectly structured data. In fact, one of AI’s more promising roles in construction is helping teams make sense of information that has historically been difficult to organize.
Capital projects generate enormous amounts of unstructured data: photos, field reports, documents, schedules, drawings, equipment information, inspection records and written notes. Even the language used to describe the same object or condition can vary across companies, geographies and projects.
AI can help connect some of those dots. Computer vision can identify objects in photographs. Optical character recognition can capture visible text. Metadata can add information about when and where an image was taken. Natural-language tools can make those records easier to search.
Suddenly, information that technically existed all along becomes much easier to use. A team trying to understand what happened in the field may be able to find relevant photos faster. A claims team may locate supporting evidence without manually sorting through thousands of files. A project controls professional may be able to retrieve historical context that previously depended on knowing where someone stored it.
This is an important point because AI does not always need to predict the future to create value. Sometimes it just needs to make what the organization already knows easier to find.
The Best AI May Eventually Become Almost Invisible
Today, companies tend to talk about AI as a separate category: there is the project controls system, and then there is the AI tool. That distinction may not last.
Kaplanoglu expects AI to become increasingly embedded in the software and workflows construction teams already use. Over time, users may not stop to think about whether a specific capability is “AI-powered” at all. That is probably a sign of maturity.
AI becomes part of how a scheduler creates a first draft, how a cost engineer investigates an anomaly, how a project leader searches for information, and how teams validate progress or identify risk. The technology itself becomes less interesting than what it allows people to accomplish.
That should also change how companies evaluate AI investments. The objective is not to accumulate the most AI features. It is to identify where intelligence, automation and easier access to information materially improve the way projects are controlled.
Start with the work, not the AI
For construction leaders deciding where to begin, the answer is probably not an enterprise-wide directive to “use more AI.” Start with the work your teams are already doing.
Look for places where experienced people spend too much time assembling information before they can apply judgment. Look for workflows that are repetitive, slow or difficult to scale. Look at the questions project leaders struggle to answer quickly.
Strong candidates often share a few characteristics:
- Experts repeatedly perform similar preparation work.
- Useful historical data already exists.
- Building the first version takes significant time.
- Information is difficult to find or scattered across systems.
- Teams repeatedly perform the same verification steps.
- Faster insight could lead to a better project decision.
Then apply AI where it improves that process without obscuring accountability. The question is not, “Where can AI replace the project controls team?” It is, “Where can AI remove the work that prevents the project controls team from doing its best work?”
That is a much more productive place to start.
What AI Changes—and What It Doesn’t
The most compelling near-term vision for AI in project controls is not a project running itself. It is a scheduler starting with a useful draft instead of a blank screen, an estimator spending more time challenging assumptions and less time recreating work, and a project leader getting to the right information faster.
In the example, a scheduling exercise that may take days could potentially produce a three-hour starting point. Thousands of project records become something teams can interrogate conversationally. Field information that was difficult to retrieve becomes searchable.
None of that removes the need for expertise. It makes expertise more productive.
On large, complex capital projects, where small improvements in speed, foresight and decision quality can have outsized consequences, that may be the AI opportunity that matters most.
Learn more or request a demo today.
Frequently Asked Questions
How is AI being used in project controls? AI can help project controls teams generate preliminary schedules and estimates, analyze cost and schedule information, search project records, identify patterns and risks, and reduce repetitive preparation work. Experienced professionals still validate the outputs and decide how to act on them.
Can AI create a construction schedule? AI can help generate a preliminary construction schedule from project information such as an RFP. The schedule should still be reviewed and refined by an experienced scheduler who understands sequencing, assumptions, delivery methods and project-specific conditions.
Will AI replace project controls professionals? AI is more likely to change how project controls professionals spend their time than eliminate their roles. It can reduce manual preparation, search and analysis while increasing the importance of validation, interpretation and project-specific judgment.
Why is data quality important for AI in project controls? AI relies on the information available to it. Inconsistent naming, incomplete records, poor-quality inputs and missing context can reduce the accuracy and usefulness of AI-generated outputs.
What is the biggest near-term opportunity for AI in project controls? One of the clearest opportunities is accelerating the work that happens before expert judgment. AI can help prepare a first draft, organize information or surface potential issues faster, giving project professionals more time to analyze, improve and act on the result.
Where should construction companies start with AI? Start with specific workflows where qualified people spend significant time on repetitive preparation, information retrieval or validation. Define the project outcome you want to improve, assess whether the necessary data is reliable, and keep experienced professionals accountable for reviewing AI-generated outputs.