BuildClaim
Construction Change-Order & Revenue Recovery AI
AI that helps contractors document, track, and recover revenue from construction change orders.
- Category
- SaaS
- Type
- Product
- Status
- pre-production validation
- Website
- buildclaim.vercel.app
01 · Overview
The problem and the solution
Business problem
On construction projects, changes in scope are often documented late or incompletely. Without a clear record, contractors struggle to support change orders and can leave earned revenue unrecovered.
Solution
BuildClaim organizes change-order documentation and uses AI to surface potential claims and assemble supporting evidence, so project teams can pursue what they're owed with confidence.
02 · Functional
What the product does
Key features
Change-order tracking
Track each change from first notice through resolution.
Evidence organization
Keep documents, photos, and correspondence tied to each change.
AI-assisted claim identification
Flag scope changes that may support a change order.
Claim package preparation
Assemble supporting documentation into a reviewable package.
User workflows
- Project documents — contracts, drawings, RFIs, and correspondence — are uploaded and indexed.
- AI reviews new documents and flags potential changed work.
- A change event is created to hold the issue, evidence, owners, and follow-up.
- Labor, material, and equipment are priced against the contractor's cost library.
- Notices and letters are drafted from the contract's requirements and deadlines.
- The change order is submitted and tracked through approval.
03 · Technical
How it's built
Technology stack
- Next.js
- TypeScript
- PostgreSQL
- OpenAI
- RAG
- LangGraph
- AWS
- Java
- Spring Boot
- React
- Docker
- Redis
- Agentic AI
System architecture
BuildClaim combines a document pipeline with a change-management workspace. Uploaded documents are parsed, split into sections, and embedded so they can be searched by meaning. A retrieval-augmented generation (RAG) layer grounds every AI suggestion in the source text and links back to where it came from.
The workspace — projects, change events, estimates, and change orders — is a Next.js application on PostgreSQL, with background workers handling document processing and an agent workflow orchestrating review steps.
Engineering highlights
- RAG pipeline with citations, so every flagged change links to its source document.
- Agent workflow (LangGraph) that classifies documents, extracts potential changes, and drafts follow-up.
- Estimating that prices line items from a contractor-maintained cost library.
- Contract-aware notice tracking that turns contract clauses into deadlines.
Challenges and design decisions
- Trust: contractors need to see why something was flagged, so the system shows evidence and citations, not just conclusions.
- Long, inconsistent documents: drawings, scanned PDFs, and email threads are normalized before analysis.
- Deadlines matter: notice periods come from the contract itself and are surfaced early instead of relying on memory.
04 · Infrastructure
Hosting, delivery, and operations
- The web app and document-processing workers run as containers on AWS; uploaded documents are stored in S3 and encrypted at rest.
- Managed PostgreSQL holds project data and vector embeddings, with automated backups and point-in-time recovery.
- Ingestion is asynchronous: uploads trigger background jobs for parsing, chunking, and embedding, with retries for large files.
- CI/CD runs checks and migrations and deploys to staging before production.
- Monitoring covers job throughput, failed documents, and LLM latency and cost per project.
- Each company's data is isolated, and access to documents is role-based.
Interested in BuildClaim?
Visit the product website, or get in touch for a walkthrough or a technical discussion.