Enterprise
SmartFACT is built for environments where rules are complex, systems are critical, and auditability is non-negotiable.
Multi-Language Rule Authoring
Domain experts author rules in their own language. German process engineers write in German. English-speaking analysts write in English. The compiler handles both with the same field catalog and produces identical IL. The English restatement provides a common audit trail regardless of input language.
Native System Execution
Compiled rules execute inside the systems they serve - no middleware hops, no external API calls at decision time, no runtime dependencies on the compilation service. The evaluator runs alongside your existing processes, whether that's inside an ERP kernel, a database procedure, a Java service, or a .NET application.
Immutable Audit Trail
Every rule compilation is stored as a versioned, immutable artefact: the original natural language input, the AI model used, the confidence score, the compiled AST, every lowered output, and the test fixture results. Compliance teams can trace any production rule back to its source.
Fail-Closed Security Model
The evaluation engine is a strict tokeniser and recursive-descent parser. No dynamic SQL, no dynamic table names, no eval(). Unknown fields return false and log an error. The system never silently returns true for something it cannot resolve. Designed for GxP and SOX-regulated environments.
Verified Before Deployment
AI-generated test fixtures exercise every compiled rule against synthetic data before deployment. The CALIBRE evaluator confirms boolean correctness. Mismatches are surfaced immediately with the fixture data visible for inspection. No rule deploys unverified.
Scale
Rules compiled by SmartFACT have been evaluated against 500,000+ records in under a minute. The hybrid database + in-memory approach pushes set-based conditions into the query layer and handles complex navigation at the evaluator level. Production-ready performance on real enterprise data volumes.
Compile Once, Execute Anywhere
The SmartFACT compiler produces a single typed AST from natural language. Target-specific lowerers translate it into the form each system can execute natively. Your authoring experience stays consistent; your execution runtime matches the system.
SQL WHERE Clauses
Lower to parameterised SQL predicates for PostgreSQL, Oracle, SQL Server, MaxDB, and HANA. Execute directly against your data warehouse or operational database, with set-based filtering pushed down to the engine.
OData $filter
Generate OData 4.01 filter expressions for SAP Gateway, Microsoft Dynamics, Azure Data Services, and any system exposing an OData endpoint. Navigation, lambda quantifiers, and $count expressions supported.
JSONLogic
Produce portable JSONLogic expressions for in-memory evaluation across any language. Ship rules to edge runtimes, browser clients, or serverless functions without a database round-trip.
ABAP Function Modules
Evaluate rules natively inside SAP via fail-closed ABAP function modules. No RFC calls at decision time, no middleware latency. Runs in the same work process as your existing jobs and user exits.
LINQ & Expression Trees
Target .NET applications with LINQ expressions or compiled expression trees. Rules integrate directly into Entity Framework queries, business logic layers, and enterprise service buses.
JPA Criteria & Drools
Bridge to Java stacks with JPA Criteria API for database queries or Drools rule format for existing BRMS environments. Migrate legacy rule bases into natural language authoring without disrupting runtime.
Elasticsearch DSL
Lower to Elasticsearch query DSL for document search, log analytics, and full-text filtering. Turn natural language investigation queries into precise structured searches across your indexed data.
GraphQL Filters
Generate filter variables for GraphQL APIs. Rules authored once by business users drive consistent filtering across web, mobile, and internal tooling - all hitting the same GraphQL layer.
Custom Targets
The AST is the single source of truth. New lowering targets are added by implementing a tree walker - no changes to authoring, catalogs, or verification. If your system speaks a query language, we can target it.
How It Fits Your Organisation
Domain Experts
Author rules in natural language using the web-based rule authoring tool. No code, no training, no IT tickets.
Developers
Integrate the compilation API and deploy the evaluation runtime in your target systems. REST in, native execution out.
Compliance
Review compiled rules, test fixtures, and confidence scores through the admin interface. Full traceability from intent to execution.
Let's Talk Enterprise
We work directly with enterprise teams to define field catalogs, validate compilation accuracy, and deploy evaluation runtimes inside the systems that matter - ERPs, databases, custom applications, and bespoke platforms.