Execution Controls
Adapter translating target records to agnostic models
Reasoning node assessing targets without platform code
Entity instance, risk record, or audit response set
Agnostic Core Processing Pipeline
// Select a GRC platform, agent, and target above to begin mapping pipeline...
// Raw database records parsed into standard schema-agnostic Zod objects...
// Prompt generated using localized guides, rubrics, and active issue contexts...
// Logs from platform-specific adapter operations writing results to database...
Live Discovery (Universal Schema Discovery Agent)
Connects directly to a live Salesforce org, introspects its real object schema, vector-matches tables/fields against our agnostic models, confirms the mapping with the LLM, then validates against live sample records — producing a config the 3 reasoning agents can run against immediately, with zero new adapter code.
Add New GRC Tool Integration (Paste Mode)
Paste a raw API response, database table schema, or unstructured documentation from a target GRC tool. The Schema Discovery Agent will dynamically create field mapping models to match our platform-agnostic models.
Mapped Adapter Schema Configuration
// Field mappings and structural alignments will output here...
AI Observability
Every agent execution is recorded as a trace: each LLM call, embedding call, and platform query/write inside it becomes a span with latency, outcome, and any self-healing actions taken. Data persists in backend/data/traces.jsonl.
Recent Traces
No traces yet — run an agent from the Sandbox tab, then hit Refresh.
Platform Agnostic GRC Architecture
By decoupling the GRC platform adapters, reasoning engine, and LLM interfaces, you can plug in any platform API or switch LLM providers without touching the core validation and compliance logic.
ServiceNow GRC
GlideRecord & REST Tables
Salesforce GRC
SOQL & Custom Objects
Future Platform
Archer / LogicGate / Jira GRC
Core GRC Reasoning Engine
TypeScript + Zod validation models
LLM Adapter Layer
Google Gemini Core (JSON Schema forced)
🛡️ Zero-Access Core
Core logic works solely on unified schema structures. Platform bugs, table modifications, or access control failures stay contained within the adapter layer.
⚡ Cost & Speed Optimization
Calculates database fingerprints from active evidence and queries prior closed audits to reuse answers. Identical states carry forward instantly with zero LLM API costs.
🔌 Instant Onboarding
Using the Schema Discovery Agent, entering raw GRC API properties generates instant translation bindings for new platforms in seconds.