Plain-language definitions for the terms that show up in construction technology, AI, and lifecycle operations.
Buyers and operators evaluate construction technology faster when the vocabulary is shared. A clear glossary supports that.
A–Z anchors below. Cursor wires scroll behavior on copy-back.
A class of software where AI is the substrate, not a feature. Every entity (drawing, RFI, PO, timesheet, inspection) lives on one project graph; AI reasons across the whole lifecycle to recommend next actions — not just summarize what already happened.
Replaces siloed module databases. The project graph is what allows AI to answer cross-functional questions: "Which RFIs are causing schedule slip on jobs over $20M?" requires data from RFIs, schedule and finance — answered in one query, not three exports.
AI-native systems were architected so AI could reason across all data without integration glue. AI-bolted-on systems retrofit a chatbot panel onto legacy modules; the AI sees one module at a time and hallucinates across the gaps.
Demos well, ships poorly. Bolted-on AI cannot reason across modules because the modules do not share a model. Distinguishable in a demo by asking for an answer that requires three modules at once.
The wedge against pre-construction OR ops point tools. Lifecycle intelligence makes preconstruction estimates inform field decisions and lets field reality re-tune the next bid. Data is not re-keyed across phases.
Reports tell you the schedule slipped 6%. Decision support tells you which 3 actions recover the float, the historical success rate of each, and the projected cash impact.
A queryable archive of decisions, lessons learned and outcomes from every project the org ever ran. AI uses it to ground recommendations in your own historical performance — not generic best practice.
A structural product choice. Each role gets its own context window, access scope, metric set and escalation thresholds. The same question — "how is project X going?" — has at least four right answers depending on who asked.
Quantified loss from swivel-chair work, double entry and tool-to-tool reconciliation. Mid-market GCs measure it at 6+ hours per PM per week, plus an invisible AI gap (bolted-on AI hallucinates because the data underneath is fragmented).
EstimateGPT, ScheduleGPT, PayrollAI, QualityGPT, ToolboxTalksAI. Purpose-built rather than generic. Each agent has its own scope, its own evaluation harness, and its own role-tuned outputs.
A trust default. Construction project data is sensitive — bid sheets, contracts, payroll, site photos. Zero-retention AI means the LLM provider does not retain or learn from any of it. Required for enterprise procurement.
SOC 2 Type II, GDPR, regional data residency, SSO, SCIM, immutable audit log, scoped access, zero-retention AI. Not a checklist on a security page — wired into every workflow.
A live, AI-projected number — not a month-end report. Combines unbilled work, contingency burn, change-order drift, and forecasted cost-to-complete into one figure per project and per portfolio.
The single best leading indicator of schedule slip on a build. AI-native ops shrinks this by drafting RFIs from voice/photo, routing them to the right reviewer, and surfacing similar past RFIs as precedent.
Computed from current pace, dependency depth, weather, RFI backlog, and historical slip on lookalike milestones. Reviewed weekly in portfolio standup; the top five are the only schedule conversations worth having.
AI watches every cost code, every change order, every productivity datapoint and reports the delta vs the as-sold margin in near real time. Lets the PM intervene while the project is still recoverable.
Lookalike retrieval from your project graph + assembly drafting + win-probability scoring. Estimators bid more jobs, sharper, with the org's actual production rates baked in — not generic R.S. Means.
Each handoff is a chance for drift. AI-native ops collapses them: a voice note in the field becomes the daily log, the RFI draft, the safety record and the photo metadata in a single capture.
When cost, billing, and field actuals share one graph, the close is a query, not a project. Finance reviews variances daily; nobody waits 30 days to learn what went wrong.
AI scans the network for parallelisable activities, vendor lead-time slack, and resource swaps that historically worked on similar jobs. Recommends the cheapest 3 moves to recover X days.
A function of reviewer load, completeness of package, and AI pre-screening. Slow submittals starve the schedule; fast ones unlock procurement and prevent slip.
AI tags punch items with predicted complete-by dates from history and alerts the super when decay accelerates. The closeout phase becomes a managed flow, not a final-week panic.
Tracked live and per-project. AI flags when drift on any project crosses an org-defined threshold — long before the variance report would have caught it.
Predicts dispute outcomes, claim defensibility and field-office handoff health. AI scores logs as they're submitted and nudges the super when quality drops.
A glossary entry rarely stands alone. Most terms belong to a broader operating concept used across the lifecycle.
Static preview of the related-intelligence buckets a real resolver will populate.
Tell us about the workflow. We will walk through the supporting intelligence — no scripted demo.