Capture → Knowledge Graph → Attributed Answer
SpaceCapture AI turns field captures into verifiable, structured data: scan a space, resolve it into a knowledge graph, and get answers with source attribution — not guesses.
The Capture-to-Answer Pipeline
Three steps from the field to a citable, auditable AI answer.
1. Capture
Inspectors capture photos, video, 360° reality capture, and LiDAR scans from any mobile device — fully offline. Every image is tagged with project, location, and timestamp metadata at the point of capture.
2. Knowledge Graph
Captured evidence is ingested into a structured knowledge graph. Assets, defects, locations, and inspections become linked entities, so the site becomes queryable structured data instead of scattered files.
3. Attributed Answer
AI answers are grounded by VeritasGraph with source attribution and a clear reasoning path — every claim links back to the exact capture or record it came from, so answers are verifiable, not guessed.
Search and AI answers on SpaceCapture are grounded by the VeritasGraph open-source GraphRAG framework. VeritasGraph links every claim to its origin capture through knowledge-graph reasoning and multi-hop traversal, so each answer ships with source attribution and a verifiable reasoning path.
Verifiable AI with source attribution · GraphRAG · knowledge-graph reasoning · multi-hop traversal · on-premise AI
AI Answers & Source Attribution FAQ
How SpaceCapture uses VeritasGraph to verify answers and ground them in a knowledge graph.
What is VeritasGraph and how does SpaceCapture use it?
VeritasGraph is an open-source framework that connects your project data into a knowledge graph, so every AI answer can show where it came from. SpaceCapture uses it to ground search and chat: instead of guessing from similar-sounding text, answers are built from your own captures, manuals, and reports — with the source attached.
How does source attribution work in SpaceCapture?
Every sentence the AI writes carries a link back to its origin — the photo, inspection record, or report it was built from. VeritasGraph keeps those links intact as it reasons across the graph, so you can open any claim and land on the actual evidence instead of a paraphrase.
What is a knowledge graph and why does it matter for inspections?
A knowledge graph stores your site as connections instead of folders: a defect linked to an asset, the asset linked to a floor, the floor linked to your project. Ask "which defects are still open on level 3?" and you get one precise, sourced answer — not a folder of photos to open one by one.
How does SpaceCapture verify AI answers?
Two things ship with every answer: a citation to the original capture, and the path the graph took to build it. Open either one and you're looking at the evidence itself. That's how an inspector or auditor checks a claim in seconds — by reading the source, not by trusting the bot.
What is GraphRAG and how is it different from regular AI search?
Regular AI search finds text that resembles your question and rewrites it. GraphRAG follows the relationships between your captures and records to construct the answer, then hands you citations and a visible reasoning path. The result is grounded in your data — not in whichever paragraph happened to match.
How does a photo capture become a knowledge graph?
Each photo or scan is tagged with its project, asset, and location the moment it's captured. SpaceCapture then reads the image for what matters — a crack, a rusted bolt, a severity score — and stores it as linked nodes: defect → on asset → in zone → within project.
Is VeritasGraph deployed on-premise or in the cloud?
Your choice. Run SpaceCapture on cloud hosting in the US, EU, or APAC regions, or keep the entire stack on-premise — including air-gapped setups for defense and critical-infrastructure teams. Either way, your knowledge graph and captures stay inside infrastructure you control.
Start Inspecting Smarter
with SpaceCapture
Replace paper checklists and manual photo logs with AI-powered defect detection, automated reports, and real-time site visibility — from any device.
Available on iOS, Android, and Web