Everyone is racing toward artificial intelligence, but not everyone is ready for it. The truth is you cannot automate what you do not understand, and you cannot scale what is not connected. In the architecture, engineering and construction industries, the key to AI lies in the tedious and unglamorous work that goes into structuring, governing and integrating data across complex, fragmented systems.
As I often remind my peers, AI is only as effective and impactful as the data driving it. Before you can leverage it, you must first establish the foundation. Many are distracted by “shiny new buttons” and lose sight of the importance of longevity. What is missing in most AI conversations is foundational readiness driven by data hygiene, governance and interoperability. By incorporating that into your data strategy, not only will it enable your efforts to scale sustainably, but it will also decrease resistance to innovation.
Innovation is inherently messy, but structure keeps the mess from devolving into chaos. In AEC, data lives everywhere: internal enterprise resource planning systems, project servers, web-based platforms, client- or consultant-driven external/shared systems, and emails. The goal is not to force everything into one box but rather to build an approach that connects what should be connected and accepts that some fragmentation is inevitable. Be intentional with your efforts, and leave capacity for the integration or evaluation of more agnostic technology.
Remember — digital transformation is a rhythmic process. Crawl, walk, run, but then check back in to refine as needed. As new tools emerge, taking a moment to reestablish your structure allows you to continually refine the data model. That cadence keeps teams from bolting AI solutions onto volatile foundations, ensuring data matures in alignment with your business objectives.
We learned this the hard way. Like many firms, we rolled out an ERP and declared it our “single source of truth.” It sounded great until we started onboarding and discovered various client nuances. Implementing a minimum viable product was not enough if we wanted to avoid duplicate efforts.
Then came the cascade of practical questions. How do we connect the rest? How do we make reporting easy? Where do people search? That is when we knew we needed to shift to a hybrid data strategy, consolidating where it added value and thoughtfully integrating key data sets between systems, all while maintaining continuity. The hybrid approach has opened the door to several process improvements, and we are hopeful that it will enable us to gracefully lead our clients through this era of digital innovation.
When you start ingesting years of project-server data, you will quickly uncover inconsistent fields, naming drift, duplicates and outdated records. That part of the process can be overwhelming, but it is the necessary work that goes into building a data structure. Bring a diverse team of stakeholders into the process to give different perspectives and expertise to the table. In time, the structure will reveal itself. Prioritizing diligence over speed will pay off in reliability and insight later.
Ironically, AI can be the flashlight that reveals your data gaps. Internal hackathons or AI competitions are powerful catalysts: A single use case from your day-to-day users can expose missing connections and prompt a company’s data strategy. In our case, a proof of concept guided by our employees through an internal AI competition surfaced the highest-reaching and most impactful solution, expediting buy-in to proceed with an even more comprehensive technology and data vision.
Technology only succeeds when people buy in. Data work is cultural as much as technical: Involve end users and clients early, design around their questions, and let their workflows guide the information model. In an employee stock-ownership plan culture, ownership mindsets amplify this. Every voice counts, and that inclusivity raises the quality of both taxonomy and the tools.
End users want one place to ask all the questions. AI can help deliver that but only after you connect the pieces and establish trustworthy pathways from source to surface. Otherwise, you are just automating confusion. The structure you build now is precisely what turns experimentation into durable capability.
It is never too late to start or to reassess your direction. Growth is an ongoing, often imperfect rhythm, but every step forward counts.
My advice? Embrace the mess, but build a solid foundation. Connect what matters, accept strategic fragmentation, and continually refine the model to strengthen it. Do that, and you will be ready not just to implement AI but to scale it with confidence.
Brittany Pylant is vice president of technology and digital innovation at HFA Architecture + Engineering in Bentonville. She can be reached at brittany.pylant@hfa-ae.com.
READ ALSO: LR’s MAHG Acquires 2 Health Care Industry Firms
