Construction Project Controls Copilot
Outcome
Give the construction project manager one grounded control surface across schedule, cost, RFIs, submittals and change orders so schedule slip and cost exposure are explained by the contract documents and the daily field record rather than reconstructed in a spreadsheet before each owner meeting.
| Portfolio attribute | Value |
|---|---|
| Primary persona | Project manager |
| Primary aggregate | ConstructionProject |
| Cosmos DB containers | 1 |
| Grounding data | schedules, budgets, RFIs, logs, change orders |
| Development estimate | USD 60-220 monthly dev estimate |
| Runtime options | Python FastAPI and C# ASP.NET Core |
Productivity hypothesis
These measures reflect the workflow, evidence, decision boundary, and industry risk for this use case.
| Impact type | Measure | Conservative internal assumption | Customer measurement |
|---|---|---|---|
| Productivity | Controls analysis cycle time | 10-18% reduction | (baseline median controls-analysis minutes - assisted median minutes) / baseline median minutes * 100 |
| Quality or risk | First-pass acceptance | 5-10 percentage-point improvement | assisted percentage accepted without material rework - baseline percentage accepted without material rework |
Important: Conservative internal pilot assumptions only; not public benchmarks, commitments, or expected customer results. Validate against a comparable baseline or matched control cohort with mandatory human review.
Rationale: Schedules, budgets, RFIs, logs, and changes support faster status and impact-analysis preparation.
Pilot guardrail: Keep budget, schedule, RFI, change-order, and project-control decisions human-owned.
Customer validation plan
- Baseline the current workflow with representative users and tasks.
- Run the accelerator on the same task mix with human review retained.
- Measure controls analysis cycle time and first-pass acceptance with the documented formulas.
- Replace internal assumptions with customer-observed results before a business case.
Why Azure Cosmos DB
This AI solution needs an operational data plane, not only a vector index:
- Hybrid vector and full-text retrieval combines semantic recall with exact identifiers, codes, names, and domain language.
- Hierarchical partition keys and metadata predicates keep tenant and domain boundaries in the normal query path.
- Operational JSON, source chunks, workflow and agent state, feedback, and audit records remain together for low-latency reads and governed writes.
- Change feed reacts to source and operational updates without polling.
- Autoscale throughput, configurable consistency, and multi-region distribution support variable demand and enterprise availability.
Domain workflow

| Stage | Actor | Input → output | Approval |
|---|---|---|---|
| 1. Schedule and cost update ingestion | Scheduling and ERP connectors | ScheduleActivity → periodControlSnapshot | Policy controlled |
| 2. Slip and variance attribution | Construction project controls copilot | periodControlSnapshot → driverAttributionSet | Policy controlled |
| 3. Contract entitlement grounding | Retrieval orchestrator | driverAttributionSet → ContractDocChunk | Policy controlled |
| 4. Owner narrative drafting | Construction project controls copilot | ContractDocChunk → ControlsNarrative | Policy controlled |
| 5. Project manager approval and change order submission | Project manager | ControlsNarrative → submittedChangeOrder | Human required |
Technical architecture

Cosmos DB data model
| Container | Hierarchical partition key | Documents | Search paths |
|---|---|---|---|
constructionProjects | /tenantId, /projectId | Project, ScheduleActivity, CostCommitment, ContractDocChunk, ProjectIssue, ControlsNarrative | /chunkVectorEmbedding, /subjectVectorEmbedding, /narrativeVectorEmbedding, /chunkVector, /subjectVector, /narrativeVector |
Query contract
| Query | Container | Scope | Purpose |
|---|---|---|---|
| CriticalPathSlippage | constructionProjects | single-partition | Identify critical-path activities that lost float in the latest schedule update. |
| CostCodeVariance | constructionProjects | single-partition | Show cost codes whose forecast at completion exceeds budget beyond the reporting tolerance. |
| OverdueBallInCourtIssues | constructionProjects | single-partition | List RFIs and submittals past their response due date by responsible party. |
| EntitlementClauseRetrieval | constructionProjects | single-partition | Retrieve the contract and specification language governing a claimed time extension or extra cost. |
Python and C# services route each document type to its declared container and hierarchical partition key. ETag-based writes protect mutable aggregates, while append-only records and bounded cross-partition queries preserve scale.
Azure deployment

Clone the source repository, authenticate with Azure, and run:
Set-Location accelerators/real-estate-construction/construction-project-controls-copilot
az login
./Deploy.ps1 -Runtime python -Location eastus2
Change python to csharp to deploy the .NET implementation.
Explore the implementation
- Launch signed-in demo
- Implementation article
- Domain data model
- Workflow and API contract
- Python application
- C# application
- Domain container Bicep
- Platform Bicep
- Cost and configuration
- Source repository