~0.25 hours (≈15 minutes) per step for SubHub AI.
It’s not that SubHub needs 15 minutes to perform each task — the system automates them in the background. The 0.25 hours/user step represents the light human interaction needed at each stage:
Why ~0.25 hrs per SubHub step
You scan the plan set SubHub displays and click the relevant sheets. For a 200-unit multifamily, this is a quick check — about 10–15 minutes. Room/Space Detection (Automated) SubHub does it all, but you may skim a sample of rooms to confirm boundaries closed properly. That’s a spot-check, not full review → ~10–15 minutes. Finish/Scope Mapping (Automated) The model reads schedules and applies finishes automatically. User simply confirms that materials are being recognized correctly (e.g., one quick look at a legend vs applied rooms). Transitions & Seams (Automated) SubHub marks transitions and generates seam layouts. You glance at a few corridors or large rooms to verify logic. Doors/Thresholds (Automated) System excludes openings and counts thresholds. User skims through 1–2 pages to sanity check. Measurements & Deliverables (Automated) Excel + marked PDFs generated automatically. You open and review totals, compare to gross SF of the building — a 10–15 min final check. Why it’s consistent across steps
Each step requires only a sanity check (5–15 minutes), because:
The heavy work is system-side (overnight processing). User load doesn’t scale linearly with project size — whether 50 units or 200 units, you’re still doing spot-checks, not full re-tracing. The small time allocation ensures you don’t skip review but also reflects the fact that you’re not actively clicking through thousands of rooms. ⚖️ So:
The ~0.25 hrs per feature step is shorthand for: “user reviews, spot-checks, and confirms AI output before moving on.”
That’s how we land at ~1–1.25 hours total user time for a 200-unit multifamily flooring takeoff in SubHub AI.
Sanity Checks