October 12, 2026
by Arnie Gess, president of Construction Links Network
Artificial intelligence is moving quickly into construction. Project teams are beginning to use AI to search documents, review drawings, summarize information, analyze schedules, identify risks and automate administrative workflows.
But there is a problem that receives far less attention than the technology itself.
AI depends on data.
Construction projects generate enormous amounts of information across drawings, specifications, BIM models, schedules, estimates, contracts, RFIs, submittals, field reports, photographs, safety records and financial systems. Yet much of that information remains scattered across different platforms, spreadsheets, shared drives, email inboxes and project archives.
That may not prevent a person from finding the information they need.
For AI, however, fragmented, inconsistent or outdated information can become a significant limitation.
As construction organizations move from experimenting with AI toward integrating it into real project workflows, an important question is emerging:
Is your construction data actually ready for AI?
The construction industry does not suffer from a lack of information.
A single major project can generate thousands of drawings, revisions, photographs, RFIs, submittals, inspection records, schedule updates, cost reports and other documents during its lifecycle.
The challenge is whether that information can be reliably found, understood and connected.
A project team might have current drawings inside a construction management platform, older versions on a shared drive, schedules maintained in separate software, procurement information inside spreadsheets and important decisions buried inside email conversations.
People often learn how to navigate those systems.
AI does not have that institutional knowledge unless the information is accessible and properly connected.
For construction data to become genuinely useful for AI, organizations increasingly need to think about whether their information is complete, consistent, structured, accessible and governed.
The question is therefore not simply whether a construction organization has data.
It is whether the organization can trust the data AI is being asked to use.
Poor information has always created problems in construction.
An outdated drawing can cause rework. An incomplete schedule can distort planning. Incorrect cost coding can affect forecasting. Missing procurement information can hide a potential delay.
AI can potentially make it easier to identify some of these problems.
But AI can also amplify them.
Imagine an AI system being asked to identify schedule risk while working from an outdated schedule. Or an AI assistant answering a technical question using a superseded drawing. An automated workflow could potentially take the problem further by using unreliable information to prepare or initiate the next action.
That becomes increasingly important as construction AI evolves from answering questions toward participating in workflows.
The more responsibility software is given to interpret information, recommend actions or automate tasks, the more important the quality of the underlying information becomes.
AI does not eliminate the need for reliable construction data.
It makes reliable data more important.
Construction information rarely lives in one place.
Estimating teams may use one platform. Project managers another. Accounting departments may rely on an ERP system. Field teams may use mobile applications. Designers work with BIM and design platforms. Subcontractors and suppliers may operate through their own systems.
Then there are spreadsheets, shared drives, email and decades of archived project information.
Each system may work perfectly well for its intended purpose.
The difficulty appears when AI needs to understand information across several of them.
An AI system attempting to analyze project performance, for example, might need schedule information from one source, cost data from another, field progress from another and historical information from previous projects.
If those systems describe projects, activities, companies, materials or costs differently, connecting the information becomes difficult.
That is why the next stage of construction’s digital transformation may involve more than adopting additional software.
It may require better connecting the information the industry already has.
This challenge becomes even more important with the emergence of agentic AI.
Traditional AI tools generally respond to requests. AI agents are being developed to go further by gathering information, reasoning through multiple steps and taking permitted actions within workflows.
Construction technology companies are already developing AI agents designed to assist with activities such as RFIs, submittals, project searches, daily logs and document review.
But an AI agent can only act intelligently if it has reliable context.
Consider an agent helping prepare an RFI. It may need to understand the latest drawing revision, the applicable specification, previous correspondence, related submittals and perhaps the schedule activity affected by the issue.
If those records are incomplete, disconnected or outdated, automation does not solve the problem.
It can accelerate it.
Before construction organizations ask how much work AI agents can perform, they may need to ask a more fundamental question:
What information are we allowing them to work from?
There is unlikely to be a single definition that applies to every contractor, consultant or project.
But several characteristics are becoming increasingly important. Industry research increasingly points to data quality, accessibility and governance as foundational requirements for scaling AI beyond isolated construction use cases.
AI-ready construction data should be accurate enough to trust, current enough to reflect project reality and accessible to the systems authorized to use it.
Information should also be organized consistently. Naming conventions, cost codes, project classifications and document structures become much more valuable when they are standardized across projects.
Connections matter as well.
A drawing may be useful by itself. It becomes considerably more useful when related information—such as specifications, RFIs, submittals, field observations and schedule activities—can be connected to it.
Historical information may become especially valuable. Years of estimates, schedules, productivity records, project outcomes and lessons learned could eventually help AI identify patterns that would be difficult to recognize within a single project.
But only if that history is sufficiently complete and consistent to be trusted.
Technology alone cannot make construction data AI-ready.
People create much of the information AI will eventually depend on.
Superintendents complete field reports. Project managers update schedules. Estimators classify costs. Engineers revise drawings. Coordinators manage RFIs and submittals. Accounting teams maintain financial records.
If information is captured inconsistently—or not captured at all—the problem eventually becomes part of the dataset.
That means AI readiness is partly a workforce issue.
Organizations may need clearer information standards, better training and greater consistency in how project teams capture and manage data.
This does not mean every construction professional needs to become a data scientist.
It means project teams increasingly need to understand that the information they create today may influence the decisions AI helps make tomorrow.
There is another question beyond whether construction data is accurate.
Who should be allowed to use it—and for what?
Construction organizations manage commercially sensitive information, employee records, contracts, financial information, proprietary designs, safety documentation and confidential client data.
Connecting AI to project information therefore requires clear rules and accountability.
Organizations need to understand which systems can access particular datasets, what information can be used by AI tools, where that information is stored, how outputs are verified and which decisions still require human approval.
These controls become particularly important as AI moves deeper into operational workflows.
The goal should not simply be to give AI access to more information.
It should be to give authorized AI systems access to the right information, for the right purpose, with appropriate human oversight.
Becoming AI-ready does not necessarily require replacing or rebuilding an organization’s entire technology environment.
A more practical starting point may be understanding what already exists.
Where does project information live? Which systems contain the most reliable records? Where are duplicate versions being created? Which processes still depend heavily on spreadsheets or email? Are naming conventions consistent between projects? Can historical project information actually be retrieved and compared?
Those questions can expose weaknesses long before an expensive AI implementation does.
Organizations can then begin improving the foundations: standardizing information, connecting systems where practical, clarifying ownership, improving data quality and establishing rules for AI access.
The objective is not perfect data.
Construction projects are too dynamic for that.
The objective is information reliable enough that people—and increasingly AI systems—can make better decisions from it.
For the past several years, much of the construction industry’s AI conversation has focused on what the technology can do.
Can AI review drawings?
Can it predict schedule risk?
Can it automate documentation?
Can AI agents coordinate construction workflows?
Those questions remain important.
But as the technology moves closer to everyday project operations, another question may become just as important:
Can AI trust the information we are asking it to use?
Construction companies have spent years creating enormous digital records of how projects are estimated, designed, procured, built and delivered.
That information could become one of the industry’s most valuable assets in the AI era.
But only if it can be found, understood, connected and trusted.
The construction organizations that benefit most from AI may therefore not simply be those that adopt the newest tools first.
They may be the ones that prepare their information—and their people—to use those tools well.
