Enterprise GrafDrive

Organize your corporate files into an AI-queryable knowledge graph.

Graffold's GrafDrive crawls filesystems, extracts metadata and table schemas, detects likely duplicates and versions, and gives data teams a reviewable graph of how datasets, owners, folders, and business concepts relate. The resulting knowledge graph is queryable and transparently traceable.

Which supplier file is current?

Dozens of versions across regions with no single source of truth.

From folder tree to harmonization report

Connect → Crawl → Extract → Graph → Review — then query and trace every answer back to its source.

1

Connect

Import a bounded folder tree or connect a cloud drive.

2

Crawl

Extract metadata, permissions, and folder structure.

3

Extract

Parse table schemas from CSV/XLSX and document context from PDFs.

4

Graph

Build a graph of files, datasets, columns, owners, and concepts.

5

Review

Approve mappings, resolve duplicates, export harmonization report.

① Source: Corporate Drive
📁 Corporate/
📁 Suppliers/
📁 EMEA/
📄 supplier_master_v3.xlsx
📄 supplier_master_v2_OLD.xlsx
📄 supplier_list_backup(1).csv
📁 APAC/
📄 vendors_2024_final_FINAL.csv
📄 vendor_codes_KT.xlsx
📁 NA/
📄 procurement_suppliers.csv
📄 Copy of suppliers (2).xlsx
📁 Shared/
📄 master_supplier_table.xlsx
⚠ 47 files · 4 regions · no single source of truth
②③ Extracted Metadata
supplier_master_v3.xlsx · 847 rows · J. Smith
supplier_id company_name region onboard_date
vendors_2024_final_FINAL.csv · 312 rows · K. Tanaka
vendor_code vendor_name country reg_date
Detected relationships:
0.91 v3.xlsx ↔ vendors_2024.csv
0.87 v3.xlsx ↔ procurement.csv
④⑤ Knowledge Graph
Dataset: supplier_master
supplier_id → vendor_code
company_name → vendor_name
Rel: likely_duplicate
Owner: J. Smith (EMEA)
Owner: K. Tanaka (APAC)
Canonical: supplier_master
✓ 12 families · 89 columns · 3 canonical tables
✓ Queryable · Traceable · Human-reviewed

Example: Supplier Harmonization

supplier-harmonization
1. Find every supplier onboarding dataset across the company.
2. Group likely duplicates by schema similarity.
3. Compare column schemas across EMEA, APAC, and North America.
4. Identify current versions by modified date and owner.
5. Propose a canonical supplier table with column mappings.
6. Export the evidence report with source citations.
Result: 47 files → 12 dataset families → 3 canonical tables proposed
Evidence: 94 source citations across 6 business units

Built for enterprise data teams

Read-only, permission-aware, and human-reviewed — designed for the teams responsible for governing enterprise data.

Read-only source crawl

Graffold never modifies source files.

Permission-aware views

Graph respects source access controls.

No source file mutation

Original files remain untouched.

Every proposal has evidence

Relationships cite source documents.

Human-approved mappings

No canonical assertion without review.

Audit trail for decisions

Every approval/rejection is logged.

Data Governance

Map and reconcile scattered datasets across business units.

Data Platform

Discover file-based data assets outside the warehouse.

Analytics Engineering

Find canonical sources before building new pipelines.

Enterprise Architecture

Visualize how data flows through folder structures.

Operations Analytics

Identify duplicate reports and outdated spreadsheets.

AI Readiness

Assess data landscape before training or fine-tuning.

Get a GrafDrive Audit →