The Daily Dig
Buildots has launched the Buildots Intelligence Lab, a research hub giving the construction industry free access to benchmarking data. The data is drawn from aggregated, anonymized projects worldwide. The company calls it construction's first AI-powered research hub, and it operates as an autonomous research unit within Buildots. The goal is to answer questions the industry has rarely had solid data to answer, like how a project stacks up against comparable work, or what a realistic weekly MEP output actually looks like.
Roy Danon, co-founder and CEO of Buildots, said construction has never really had a reliable source of macro-level truth to guide major decisions. He pointed to that gap as a major factor behind stagnating productivity across the industry. The Lab is Buildots' answer: giving contractors real numbers on output, trade productivity, and scheduling instead of relying on estimates.
The early findings target some persistent pain points. Data center construction shows a 20-50% gap between planned weekly MEP output and what crews actually deliver. Separately, the Lab found that the final 20% of an activity takes 27% of its total duration. Buildots describes this as a structural pattern rather than an incidental one.
The research also found that top MEP teams can work up to three times faster than the average. That points to significant untapped capacity sitting in schedule-critical work. Schedule adherence data varied widely by sector. Healthcare projects led with average adherence around 65%, followed by data centers at roughly 57%.
Commercial and industrial work landed in the low-to-mid 40% range, while education projects came in lowest, under 39%. Buildots said the Lab will keep publishing research on a regular basis, and it's inviting academics, analysts, media, and construction professionals to submit their own hypotheses and questions.
Snapshot:
Organization: Buildots
Initiative: Buildots Intelligence Lab
Launch Date: June 25, 2026
Type: AI-powered construction research hub
Structure: Operates as an autonomous research unit within Buildots
Data Source: Aggregated, anonymized data from global projects
Access: Freely available to the public
Research Pillars: Metrics, Benchmarks, Insights
Key Finding, Data Centers: 20-50% gap between planned and actual weekly MEP output
Key Finding, Activity Duration: Final 20% of an activity takes 27% of total duration
Key Finding, MEP Productivity: Top MEP teams work up to 3x faster than average
Schedule Adherence, Healthcare: ~65%
Schedule Adherence, Data Centers: ~57%
Schedule Adherence, Commercial/Industrial: Low-to-mid 40%
Schedule Adherence, Education: Under 39%
Notable Buildots Clients: Turner Construction, JE Dunn, Intel, HOCHTIEF, Bouygues
URL: buildots.com/lab
TheJobWalk Thoughts
The data center MEP gap deserves attention on procurement and staffing calls, not just scheduling. A 20-50% shortfall between planned and delivered output can stem from design changes, coordination breakdowns, staffing shortages, or procurement delays, not just an overly optimistic schedule. GCs relying on planned MEP output alone are working with an incomplete picture, and it's worth pressure-testing those numbers before they get baked into a master schedule.
The finding that the final 20% of an activity takes 27% of its duration is a useful reference for sequencing, though it describes activity-level duration rather than overall project closeout. Supers can still apply the underlying pattern by building more buffer into the tail end of schedule-critical tasks, where work tends to stall even when the bulk of it finished on pace.
The 3x spread in MEP productivity is the most actionable number here, but it works best as a starting point. A public benchmark confirms the gap exists. It doesn't identify which subcontractor sits on which end of it. GCs still need sub-specific performance data to act on it, and subs who can produce that evidence have a real opening in a market that's now aware the spread is this wide.
