While attention gravitates toward AI in design, the strongest business value often emerges elsewhere.
Most firms expect AI to make the biggest impact in design and production, where the workload is heavy and the potential for automation is visible. Yet these stages are also where the problems are too large, too complex, and too variable for AI to deliver consistent business value today.
Meanwhile, some of the greatest pain points in our industry – areas like estimating, planning, and operations – remain underserved and full of opportunity.
Why design is a tough place for AI to start
Design is creative, subjective, and constantly changing. Every project brings new requirements, site conditions, and client expectations. While AI can assist with repetitive tasks, it struggles with context – and in design, context is everything.
This complexity makes it difficult to calculate return on investment. How do you measure the value of a tool that generates options if the final choice still depends on human intuition? How do you quantify the cost of adoption when every project team works differently?
Even when design-phase AI saves time, its ROI is elusive, its benefits uneven, and its adoption curve steep.
For many firms, AI in design has become an ongoing experiment rather than a proven business driver.
Where AI can create immediate value
If we step back, it becomes clear that other parts of the project lifecycle are better suited for meaningful AI gains.
Tasks like estimation, scheduling, QA/QC, and cost control rely on data and consistency – two things AI handles well. These areas may not be sophisticated, but they’re where firms lose margin, confidence, and time.
At Trelity, we saw this potential in our work with Archiestimate, an AI-based hour estimation assistant. By training it on historical project data and freeing our innovation group from daily production demands, we gave them the bandwidth to explore how AI could identify the causes of cost drift and estimation gaps.
The result was a 10%-15% increase in estimation accuracy, but more importantly, deeper visibility into why these variances happened. That understanding gave us the ability to prevent problems instead of reacting to them.
The method: Discovering needs and building tailored solutions
Our experience reinforced an essential truth about AI in AEC: off-the-shelf tools rarely fit perfectly. Each firm operates with its own workflows, standards, and management culture. Data lives in different systems – some structured, many not.
To make AI truly work, firms need people who know where that data lives, how it connects, and how the company actually operates. That’s where a dedicated design and technology team becomes invaluable.
This team acts as a bridge between management and innovation – not only testing new tools, but also identifying where the real pain points are, shaping solutions around them, and guiding cultural adoption. They know how projects flow through the organization, how decisions are made, and where the “hidden data” can be unlocked for learning.
Without that internal understanding, even the most advanced AI remains disconnected – a solution looking for a problem.
AI as a mirror for the organization
When managed thoughtfully, AI becomes less of a futuristic add-on and more of a mirror reflecting how a firm truly works. It shows where decisions stall, where communication breaks down, and where performance varies.
The firms that benefit most won’t be the ones using the most AI features – they’ll be the ones that align technology with their own identity, turning internal knowledge into actionable intelligence.
Turning curiosity into capability
The most successful AI strategies will start small – with curious people, real data, and clear business goals. Firms that make space for exploration, guided by teams who know how the company truly operates, will find practical wins first and larger transformation later.
AI doesn’t replace human judgment; it enhances it by revealing what we couldn’t see before.
 |
Kubilay Şahinler is a design tech leader at Trelity. Connect with him on LinkedIn. |