Tuomas Syrj\u00e4nen<\/a>\u2019s keynote at the AI in AEC 2025 conference was a wake-up call. Drawing on his extensive experience in data science, consulting, and board service in the construction sector, he challenged the audience\u2019s assumptions about the industry\u2019s future.<\/p>\n Tuomas began his journey in data science in the 1990s, founded Futurice<\/a> in 2000, and currently chairs its board.<\/p>\n Throughout his journey, he has learned that digital technologies achieve exceptional results only when combined with a redesign of processes, paradigms, and organizational culture. The same holds true for AI.<\/p>\n To kick off his talk, Tuomas showed a clip<\/a> of MythBuster Adam Savage <\/strong>being pulled in a traditional rickshaw by Boston Dynamics\u2019s Spot the Robot.<\/p>\n \u201cWe take the latest and the greatest technology, whether it\u2019s AI or robots or whatever, and we put it in front of 100-year-old process, ways of working,\u201d said Tuomas. \u201cAnd then we are wondering, why don\u2019t we get any results?\u201d<\/p>\n This tendency isn\u2019t unique to construction. Tuomas also referenced a law firm CEO who admitted their business hadn\u2019t changed in 80 years, yet they were now trying to implement AI.<\/p>\n That doesn\u2019t mean proofs of concept like Savage\u2019s stunt aren\u2019t useful. They help us understand the technology. However, real benefits arise only when we go further.<\/p>\n About a decade ago, construction companies Tuomas worked with started experimenting with technology only to realize that it didn\u2019t make much of a difference.<\/p>\n They discovered that a true paradigm shift was necessary. That meant embracing principles of industrialized construction: breaking work into smaller units, increasing clock speed, identifying new KPIs, and more.<\/p>\n Tuomas shared some success stories, including a construction firm that cut cycle time from 16 to 9 months and saved \u20ac300,000 per site per month.<\/p>\n Tuomas stated that yesterday\u2019s constraints define the way we operate today. But with AI, many of those constraints no longer apply.<\/p>\n For example, you can deliver personal service without practically any extra cost. Or you can reduce process cycle time by 90% by eliminating handovers and manual compliance checks.<\/p>\n Tuomas urged the audience to challenge the assumptions they take for granted. Ask: \u201cWhat if those constraints didn\u2019t exist? How would we run our business?\u201d<\/p>\n We need full-stack thinking to achieve positive outcomes with technology; data and technology are just the foundation. To truly benefit, you must also rethink your processes, people, paradigms, culture, strategy, and business model.<\/p>\n The ultimate test is how technology helps people thrive.<\/p>\n Tuomas acknowledged that not every attempt had been a success. For instance, in one project, frontline workers were asked to report more to headquarters using a new digital tool. The goal was to provide management with real-time insights from the field.<\/p>\n But it failed because it added work for the field teams without helping them succeed, which is a dealbreaker for any change initiative.<\/p>\n AI is only as good as the data behind it. And in construction, the lack of structured, harmonized data is a big problem. Using AI on messy data isn\u2019t efficient.<\/p>\n When we break down the construction process as in Takt production, work and material management become ten or more times more granular than traditionally. That\u2019s when AI\u2019s value really becomes evident.<\/p>\n Sometimes, it\u2019s difficult to predict what data is valuable. Tuomas shared an insightful story about a company\u2019s CFO who was frustrated because they couldn\u2019t anticipate which sites would fail financially. The monthly site reports always appeared positive until the final month, when the failure was disclosed.<\/p>\n By coincidence, they discovered that access control gate data could predict a site\u2019s success 6 to 9 months in advance. No AI required.<\/p>\n The key, Tuomas emphasized, is finding the right proxy data.<\/p>\n Tuomas shared another example of what data can reveal. He said job listings are the most honest mirror of what\u2019s happening inside organizations. Everything else is talk, marketing, or wishful thinking.<\/p>\n Tuomas had collected data from job ads in construction sector companies worldwide for this year and last. The results were revealing.<\/p>\n Of the 465,200 job listings, only 3,231 mentioned AI, representing a mere 0.7% of the total. Tuomas noted that this figure is now approximately 2% in other industries. I believe that\u2019s not high, given the hype.<\/p>\n Tuomas called on the audience to look beyond tools and start imagining what transformative change could look like:<\/p>\n Most AEC firms today are AI-using<\/strong>, not AI-native<\/strong>. Tuomas explained the difference in my short interview<\/a> with him after the keynote.<\/p>\n AI-using companies apply AI to existing processes, layering it on top of outdated systems. Think auto-drafting emails, for example.<\/p>\n AI-native firms, on the other hand, reinvent<\/strong> the process. They equip frontline teams with AI tools, eliminate non-value-adding activities, and design workflows that assume AI from the start. They don\u2019t optimize; they transform.<\/p>\n The big question is: Will established AEC companies dare to become AI-native, or will new entrants, unburdened by legacy constraints, seize the 10x opportunity?<\/p>\n<\/div>\nPutting new technology in front of a 100-year-old process<\/h2>\n
![\"-\"](\"https:\/\/aec-business.com\/wp-content\/uploads\/2025\/03\/adam-demo.jpg\")
A paradigm shift<\/h2>\n
We\u2019re shackled by yesterday\u2019s constraints<\/h2>\n
You need a full-stack view<\/h2>\n
What data?<\/h2>\n
AI in construction: The buzz does not materialize<\/h2>\n
![\"-\"](\"https:\/\/aec-business.com\/wp-content\/uploads\/2025\/03\/ai-job-listings-chart.png\")
Key takeaways from Tuomas<\/h2>\n
\n
Your choice: Be AI-using or become AI-native<\/h3>\n