Turn siloed data into connected knowledge
Fluree Sense uses AI to classify, resolve, and semantically map your structured enterprise data — then continuously keeps it synchronized. No massive ETL projects. Just clean, linked, AI-ready data flowing into your knowledge graph.
⊞ERP
customer_master
⊟CSV
product_extract
⊜API
events_stream
⊛CRM
contact_table
Customer
ex:entity
Supplier
ex:organization
Material
ex:product
Plant
ex:facility
Fluree Sense uses AI to classify source data against a canonical model, resolve cross-system entities, and create semantic links that turn disconnected records into connected knowledge — without ripping out source systems.
The Sense Pipeline
From raw sources to knowledge graph in seven stages.
Plug into the systems you already run.
Sense ships connectors for the platforms enterprises already depend on — databases, SaaS apps, warehouses, files, APIs. Onboarding a new source is configuration, not engineering.
Databases
Oracle · PostgreSQL · MySQL · SQL Server
Enterprise apps
SAP · Salesforce · Workday · ServiceNow
Warehouses & lakes
Snowflake · Databricks · BigQuery · Iceberg
Files & extracts
CSV · TSV · Fixed-width · Mainframe · Excel
APIs
REST · SOAP endpoints
Three layers of AI classification turn raw sources into modeled meaning.
Tables, columns, and records get aligned to your semantic model — even the records your source system never explicitly modeled. High-confidence decisions auto-promote; ambiguous ones route to stewards for review before they reach the governed graph.
Before
acct_nm
cust_type
cntry_cd
status_flag
After
ex:accountName
98% confidence
ex:customerType
95% confidence
ex:country
ISO IRI
ex:lifecycleStatus
classified
Standardize values so cross-system queries actually agree.
Reference values — countries, currencies, status codes, units of measure — get mapped to controlled vocabularies and canonical IRIs.
"USA"→iso3166:US
"United States"→iso3166:US
"U.S."→iso3166:US
"Active"→ex:status/active
"USD 1,250.00"→1250.00 [iso4217:USD]
One governed identity for every real-world thing.
Records describing the same customer, supplier, or product across CRM, ERP, and billing get resolved into a single Golden Record — with full provenance back to every source.
CRM
Acme Corp
crm-1188
ERP
ACME Corporation
erp-4421
Billing
Acme Co.
bill-771
Golden Record
ex:organization/acme
Lineage preserved to CRM, ERP, and billing records with attribute-level provenance.
What belongs in the graph, and what stays at the edge.
Master data and decision-critical facts persist in Fluree Core as JSON-LD; high-volume operational detail stays in source systems and remains queryable via R2RML federation. Quality scores — completeness, consistency, timeliness — attach to each fact as it’s mapped, so downstream agents see trust signals next to value.
Resolved data lands in Fluree Core as governed truth.
Triples are transacted into Core with full temporal history and lineage. Every fact carries quality metadata and source provenance, ready for AI, GraphRAG, analytics, and application use the moment it lands.
The graph stays current after onboarding.
Source updates flow through the same classification and resolution pipeline that built the original graph — no nightly batches, no manual re-imports, no lag between operational reality and what your AI sees.
Source system update detected
Record reclassified and re-resolved
Golden Record transacted into Fluree Core
Subscribed systems notified
Search 300+ sources…
AVAILABLE SOURCES
SalesforceApp · OAuth
SnowflakeData lake
PostgresDatabase · replica
customers.csvCSV · 1.24M rows
1.24M rows staged
streaming to Fluree · CONNECTED
- Connect any source.
CSV, API, Postgres, Snowflake, Salesforce — Sense ingests them as-is. No schema migration. No pipelines to maintain.
- The graph builds itself.
Entities resolve, duplicates merge, and relationships infer in place — no modeling marathon, no manual ontology.
The old way
| Capability | Traditional ETL | MDM | Fluree Sense |
|---|---|---|---|
| Discovery & Onboarding | Manual config | Manual config | AI-driven classification |
| Schema Mapping | Hand-coded transforms | Rule-based | ML-trained from SMEs |
| Entity Resolution | Separate tool needed | Built-in | Built-in, continuous, with governance APIs |
| Output Format | Tables / flat files | Tables | Semantic knowledge graph (JSON-LD + R2RML) |
| Continuous Sync (CDC) | Requires extra tooling | Limited | Native, bi-directional |
| Data Lineage | Varies by tool | Varies | End-to-end, built into graph |
| Knowledge Graph Ready | No | No | Direct load to Fluree Core |
| Human-in-the-Loop | Not typical | Some support | Built-in adjudication queues |
| Federated Query Output | No | No | R2RML virtual graph views |
Sense is the structured data on-ramp.
Sense brings structured systems — databases, SaaS, files, APIs — into the same governed semantic model that CAM, Core, and ITM share.
- Fluree ITM defines the language.
- Sense maps structured data into that model.
- CAM extracts knowledge from unstructured content against the same vocabulary.
- Core persists the resulting graph as governed enterprise truth.
- Fluree AI turns that connected knowledge into grounded assistants, retrieval experiences, and agent workflows.
Unstructured
Documents, PDFs, content repositories.
extract · tag · link
Open Fluree CAM
Structured
App databases, lakes, warehouses.
ingest · classify · resolve
Open Fluree Sense
· knowledge graph· vector embeddings· schemas · ontologies· policies · lineage
Open Core
The shared language every source, schema, and query aligns to.
Open ITM
Framework Model, Map, Connect — how Fluree builds sustainable knowledge systems
The framework behind Sense: ontology modeling, AI classification, and embedded security across every source.
Read the article
See Sense on your data.
Walk through your sources with a Fluree solutions architect.