mathias jonsson

Paper: papers.ssrn.com/sol3/papers....

A Harvard study involving 776 Procter & Gamble professionals suggests AI now functions effectively as a second human teammate.

Construction executives make two common AI mistakes:
1. Overestimating what AI can do in the next year
2. Underestimating what it will do in the next five

Success requires balancing short-term pragmatism with long-term vision.

The most powerful phrase in construction technology transformation: "Show me how you work now." Before changing anything, understand the current process, shortcuts, workarounds, and all. Optimization starts with observation.

Stop training people on software. Start embedding software in existing workflows. The difference isn't semantic, it's the gap between implementation failure and success.

A 9-word construction technology strategy that actually works: Identify frictions. Address one at a time. Measure the impact.

The RFP process hasn't fundamentally changed in 50 years, but the firms winning work consistently always take advantage of new tech.

They're now using analytics to identify requirements 10x faster, assess win probability with 90% accuracy, and focus on opportunities they're best positioned to win.

Quick tip: Before buying any new software, count how many clicks it takes to complete your team's three most common tasks. Anything over 5 clicks per task will face adoption resistance, regardless of its capabilities.

Construction firms obsess over field productivity but ignore office efficiency. Your estimating team spending 60% of their time looking for information isn't a technology problem, it's a profit leak.

The 3 non-negotiables for construction AI implementation: • Start with a specific business problem, not a technology • Ensure the solution fits into existing workflows • Measure success by user adoption, not features delivered

The future of construction isn't fully automated jobsites; it's hybrid intelligence. Humans bringing judgment, creativity, and context sensitivity while AI handles pattern recognition, prediction, and optimization. The most successful implementations are designing for collaboration, not replacement.

Construction leaders don't need more data; they need better questions. The companies seeing the greatest ROI from analytics aren't collecting more information, they're asking more precise questions that drive action. Start with the decision you need to make, then determine what data would inform it.

The real technological revolution is slow but steady universal interoperability.

The ability for different systems, different platforms, to communicate and collaborate seamlessly.

Forget walled gardens. Build bridges (APIs).

Email is the most underutilized data source in construction. The average company inbox contains years of informal knowledge, client preferences, and early warning signs of project issues, all gathering digital dust.

MLOps 101: Don't start with ML. Begin with the simplest solution. Get it working, refine it, and then optimize.

Spent way too much time tweaking models that didn’t solve the real problem.

Robots are friends

changelog.cursor.com