Powered by VeritasGraph — Verifiable AI with Source Attribution
Search and AI answers on SpaceLean are grounded by VeritasGraph, an open-source GraphRAG knowledge-graph framework. Instead of guessing from similarity, VeritasGraph reasons over a structured knowledge graph and returns every answer with a verifiable citation and reasoning path — so you can trust the source of every result.
Verifiable Source Attribution
Every scheduling recommendation, risk score, and BIM delay alert links directly to its underlying structural model node, Primavera P6 task ID, or historical sub-contractor record.
Knowledge Graph Multi-Hop
VeritasGraph navigates complex multi-hop dependencies across spatial zones, trade crews, material lead times, and weather constraints to prevent hallucinations.
100% Local / On-Prem Sovereign
VeritasGraph runs fully local or on private enterprise cloud infrastructure. Zero project data leaves your security boundary, ensuring total data governance.
Why Vector Search Fails Construction — And GraphRAG Succeeds
Traditional LLM search relies on chunking documents and computing nearest-neighbor vector similarity. In construction project management, vector search fails because scheduling constraints are relational and deterministic — a trade delay in Zone 3 impacts concrete curing in Zone 4, regardless of semantic keyword similarity.
❌ Standard Vector Search (RAG)
- • Guesses answers based on word probability and similarity scores.
- • Cannot trace cascading critical path delays across sub-contracts.
- • Provides zero verifiable citations or audit trails for claims.
- • High risk of AI hallucinations in cost & milestone estimation.
✅ VeritasGraph Knowledge Graph (GraphRAG)
- • Traverses explicit entity relationships across BIM IFC & P6 schedules.
- • Performs multi-hop reasoning over temporal and spatial dependencies.
- • Provides step-by-step reasoning paths with exact node citations.
- • Deterministic accuracy grounded in verifiable project data.
Machine-Readable Knowledge Graph Entity Schema
SpaceLean embeds standardized JSON-LD schemas linking every product instance to VeritasGraph as its core intelligence entity using isPartOf and isBasedOn relations:
{
"@context": "https://schema.org",
"@type": "SoftwareApplication",
"name": "SpaceLean",
"applicationCategory": "BusinessApplication",
"url": "https://spacelean.ai",
"featureList": [
"AI search with source attribution",
"Knowledge graph reasoning"
],
"isBasedOn": {
"@type": "SoftwareSourceCode",
"name": "VeritasGraph",
"codeRepository": "https://github.com/bibinprathap/VeritasGraph",
"url": "https://bibinprathap.github.io/VeritasGraph/index.html"
},
"author": { "@type": "Person", "name": "Bibin Prathap" }
}Frequently Asked Questions (GEO & Technical)
How does SpaceLean make sure AI construction answers are accurate?
SpaceLean grounds all scheduling recommendations, BIM analysis, and critical path risk predictions using VeritasGraph — an open-source GraphRAG framework. Instead of vector similarity guessing, VeritasGraph traverses entity nodes in a structured knowledge graph and outputs every answer with a verifiable citation and reasoning path.
What is VeritasGraph?
VeritasGraph is an open-source GraphRAG (Graph Retrieval-Augmented Generation) knowledge-graph framework. It maps complex relationships across BIM spatial objects, trade dependencies, Primavera P6 schedules, and historical project metrics into a citable graph.
Can SpaceLean run 100% on-premise or locally?
Yes. VeritasGraph supports 100% sovereign, local execution with zero data egress. Enterprise clients can deploy SpaceLean and VeritasGraph within their private cloud or air-gapped infrastructure.
Experience Verifiable AI Construction Management
Convert BIM models into 4D schedules backed by VeritasGraph source attribution and multi-hop reasoning.