SemanticX

Give Your AI aShared Language

The domain semantic layer that grounds LLMs, agents, and RAG pipelines in validated, computable business knowledge — so “revenue” always means the same thing.

4 Agent Tools|W3C Standards|AST-Safe Computation
Explore Domains

The Challenge

Same Word. Different Meaning. Wrong Answer.

When every system defines terms differently, AI confidently delivers the wrong answer.

Without SemanticX

Three systems, three meanings

Sales CRMRevenue

Gross bookings including pending deals

Finance ERPRevenue

Net recognized revenue after adjustments

BI DashboardRevenue

Annualized run-rate from monthly data

LLM Response (Ungrounded)

"Revenue is $4.2M this quarter." (Using gross bookings — overstated by 28% vs. IFRS-recognized figure.)

With SemanticX

One canonical definition, everywhere

Sales CRMRevenue

Canonical: Net recognized revenue (IFRS 15)

Finance ERPRevenue

Canonical: Net recognized revenue (IFRS 15)

BI DashboardRevenue

Canonical: Net recognized revenue (IFRS 15)

LLM Response (Grounded)

"Net recognized revenue is $3.28M this quarter (IFRS 15). Source: canonical term 'revenue' — resolved via SemanticX."

Enterprise AI fails not because models are dumb — but because they lack a shared vocabulary.

SemanticX gives every model, agent, and pipeline the same governed definition for every term.

The Architecture

Three Layers of Domain Intelligence

From canonical terms to computable formulas to mapped data sources — a complete semantic stack.

Query: “What's Gross Margin?”
Terminology

Define every business term once — with synonyms, abbreviations, relationships, and hierarchy. Fuzzy resolution ensures even informal queries map to the right concept.

Term definition + relationships
Computation

Register formulas that reference canonical terms as inputs. Execution is AST-based — no eval(), no injection risk. Every computation is validated, auditable, and reproducible.

Formula inputs + computation result
Data Semantics

Map semantic concepts to actual data sources — PostgreSQL tables, BigQuery, S3. Define entity relationships with cardinality and join keys, then explore the unified knowledge graph.

Validated answer: “Gross Margin = 33.3%”

Agent-Native Design

Built for the Agentic Era

Four MCP tools designed for LLM tool-calling. No wrappers — agents call SemanticX directly.

resolve_term

Multi-strategy term lookup: exact → synonym → abbreviation → fuzzy

resolve_term(term, domain?)
validate_term

Check if a term exists, get near-match suggestions

validate_term(term, domain?)
compute_metric

Execute a named formula with validated inputs

compute_metric(formula, inputs)
ask_semantic

Natural language query across the semantic layer

ask_semantic(question, domain?)
Agent Execution Trace
streaming...
TASK

What was our gross margin last quarter?

THINKINGfinancial_analyst

I need to resolve "gross margin" in the FSO domain to get the canonical definition and formula.

TOOL CALLresolve_term

resolve_term("gross margin", domain="fso")

RESULTresolve_term

{ name: "Gross Margin", formula: "gross_margin_pct", inputs: ["revenue", "cogs"] }

TOOL CALLcompute_metric

compute_metric("gross_margin_pct", { revenue: 2400000, cogs: 1600000 })

RESULTcompute_metric

{ result: 33.33, unit: "%", formula: "(revenue - cogs) / revenue × 100" }

RESPONSE

Gross margin was 33.3% last quarter, computed as (Revenue − COGS) ÷ Revenue × 100. Source: SemanticX FSO domain.

SemanticX turns agents from confident guessers into validated reasoners.

Every answer is grounded in canonical definitions. Every computation is auditable. Every term is traceable to its source.

Pre-Built Domains

From Financial Services to Power Generation

Pre-built domain ontologies ready to deploy — or create your own from scratch.

Financial Services

Revenue recognition, profitability metrics, and financial statement ontology following IFRS/GAAP standards.

32Terms
16Formulas
10Entities

Inventory Management

SKU classification, reorder optimization, warehouse operations, and demand forecasting terminology.

26Terms
11Formulas
8Entities

Power Generation

Generation capacity, transmission losses, load balancing, and grid operations across the energy value chain.

22Terms
13Formulas
7Entities

Singapore Tax

Corporate tax provisions, GST treatment, transfer pricing, and Singapore-specific tax compliance terminology.

21Terms
9Formulas
6Entities

Need a custom domain? Create one from scratch, import from W3C ontologies, or extract directly from PDF documentation.

Bootstrapping

From PDF to Production Domain in Minutes

Upload existing documentation. AI extracts the ontology. You review and deploy.

STEP 01

Upload

Upload domain documentation — PDF up to 50MB

STEP 02

Extract

AI extracts terms, formulas, and entities with confidence scores

STEP 03

Review

Review extracted items, adjust selections, verify accuracy

STEP 04

Deploy

One click creates a fully populated semantic domain

Enterprise Ready

W3C Standards. Full Portability.

Import from existing ontologies, export in any W3C format, and integrate with your data governance stack.

JSON-LDLinked Data interchange
SKOSW3C taxonomy standard
OWLWeb Ontology Language
CSV/ZIPTabular bulk export
Import & ExportBidirectional ontology transfer in all four formats
Merge StrategiesSkip, update, or error on conflicts — you control the merge
Governance CompatibleIntegrate with Collibra, Alation, and other data catalogs

Under the Hood

The Semantic Resolution Engine

From agent query to validated answer — resolved, computed, and audited in milliseconds.

Query InAgent Tool Call
agent query
Semantic Resolution Engine
UNDERSTAND

Intent Detection

RapidFuzz NLP
parsed intent
RESOLVE

Term Resolution

pg_trgm · RapidFuzz
Resolution Cascade
EXACT
SYN
ABBR
FUZZY
entry points
Agent Tool Registry
Resolve Term
Validate Term
Compute Metric
Ask Semantic
ResolvedTerm
COMPUTE

AST Execution

AST Parser · Safe Exec
validated result
ValidatedJSON · Structured
Infrastructure Layer
PostgreSQLSemantic schema + pg_trgm
MinIOPDF storage
LangfuseLLM tracing
OpenTelemetryRequest tracing

Ecosystem

The Connective Tissue of DecisionOS

SemanticX enriches KnowledgeX RAG with domain context, gives AgentsX validated semantic tools, and traces every call through MonitoringX for full cost and audit visibility.

Your domain, defined.Your AI, grounded.

Explore pre-built domains, upload your own documentation, or build from scratch.

W3C Standards
AST-Safe Computation
Multi-Domain Isolation
Agent-Native Tools