Decision Intelligence Platform

Operational Risk Intelligence,
End-to-End

From raw transaction data to explainable, auditable decisions — purpose-built for organizations where financial integrity, operational accountability, and regulatory compliance demand more than a black box.

3

ML Model Tiers

3

Explainability Levels

3+

Cloud Providers

No-Code Config

Three Pillars of Intelligence

End-to-end coverage from detection to decision — each pillar purpose-built for the demands of operational risk management.

Fraud & Anomaly Detection

Machine learning models that score every transaction against learned behavioral baselines — with configurable thresholds, post-processing rule overrides, and a granular feature contribution trail for every flagged record.

  • Multi-tier ensemble ML scoring
  • Rule-based post-processing override
  • Feature contribution attribution
  • Real-time & batch inference

Time-Series Forecasting

Multi-model ensemble forecasting across any time-series dimension — revenue cycles, claim frequencies, freight volumes — with probabilistic confidence intervals and trend explainability at the series and portfolio level.

  • ARIMA + ExponentialSmoothing + Theta ensemble
  • Confidence interval bands
  • Horizon-configurable predictions
  • Per-series AI explanation

Structured Case Management

Investigation workflows built for analyst accountability: review flagged transactions, record decisions, escalate cases, and maintain an immutable record of every human judgment — with full AI-context at the point of review.

  • Analyst review & decision labeling
  • AI-generated case summaries
  • Investigation workflow & escalation
  • Immutable decision audit trail

Four Studios. One Platform.

Each studio is a self-contained module. Together they form a complete operational risk intelligence pipeline — from raw data to final decision.

Data Studio

Ingest, validate, cleanse, and engineer features from any transaction schema. Configurable field mappings, cleansing rules, and derived feature expressions — no code required.

  • Schema-agnostic JSON ingestion
  • Field mapping for any TMS/ERP format
  • Configurable cleansing rules
  • Expression-based feature engineering
  • Data quality reporting

Model Studio

Train tiered ML models, manage the model registry, and configure thresholds. From local ensemble models to cloud-hosted endpoints on AWS SageMaker, Azure AI Foundry, and GCP Vertex AI.

  • Tier 1 & 2 local model training
  • Cloud provider abstraction layer
  • Model registry & versioning
  • Human-in-the-loop threshold tuning
  • Training job monitoring

Detection Studio

Explore anomaly scores, forecast curves, and transaction flow visualizations. AI explainability at three levels — individual transaction, case summary, and portfolio — powered by GPT-4o mini.

  • Per-transaction anomaly scores
  • Sankey flow visualization
  • 3-level AI explainability
  • Time-series forecast explorer
  • Batch inference pipeline

Case Management Studio

Structured investigation workflows with analyst review, decision labeling, and full case accountability. Every action timestamped and queryable.

  • Flagged transaction review queue
  • Analyst decision labeling
  • Case investigation grouping
  • AI case summaries on demand
  • Decision audit trail

Governance Built In

Explainability, human oversight, and auditability aren't features you add later — they're woven into every layer of the platform.

AI Explainability — 3 Levels

Every detection is explainable. At the transaction level: which features drove the score. At the case level: what the investigation pattern means. At the portfolio level: what today's results say about systemic risk. Powered by GPT-4o mini, grounded strictly in your data.

TransactionCasePortfolio

Human-in-the-Loop Reviews

Analysts remain in control. Every flagged transaction can be reviewed, labeled, and escalated through a structured workflow. Model outputs inform human judgment — they don't replace it. Analyst decisions feed back into the training and tuning cycle.

Review queueDecision labelingFeedback loop

Immutable Audit Trail

Every AI explanation is stored with the full prompt used to generate it. Every analyst decision is timestamped and attributed. Every model version is versioned and retrievable. This is the foundation for regulatory defensibility.

AI explanation logAnalyst decision recordModel lineage

No-Code Configuration Layer

Domain-specific feature engineering expressions, cleansing rules, anomaly thresholds, and field mappings are all configurable through the admin interface — without code changes or redeployment. Different clients, different rules, zero engineering overhead.

Feature expressionsCleansing rulesAnomaly overrides

Built for the Enterprise

A tiered, provider-agnostic architecture that scales from local models to managed cloud endpoints — without rewriting your detection logic.

Tiered Model Architecture

T1

Local · Deterministic

IsolationForest · AutoRegression

T2

Local · Ensemble

Outliers 6-detector · Darts 3-model

T3

Cloud · Managed

SageMaker · Azure AI · GCP Vertex

The provider abstraction layer means training dispatch is cloud-agnostic. Swap providers without touching detection logic.

Cloud Provider Agnostic

AWS SageMaker

Train & host RCF, DeepAR, and custom models via managed endpoints

Azure AI Foundry

Leverage Azure ML pipelines and registered model endpoints

GCP Vertex AI

Deploy to Vertex AI endpoints with AutoML and custom training

MLflow-compatible model registry. Not married to any provider — add or swap endpoints via the provider registry.

Transaction Data Pipelines

Schema-agnostic ingestion with configurable field mappings per org and domain. Cleansing and feature engineering pipelines execute from DB-stored rule definitions — no code deployment required for new clients.

Sankey Flow Visualization

Visualize transaction volumes and fund flows between domains, accounts, and channels as an interactive Sankey diagram. Instantly surfaces concentration risk, unusual routing, and domain-level volume anomalies that tabular data hides.

API-First Architecture

Fully Integrable.
Fully Operable.

Every capability in TransIQ is exposed through a documented REST API — built to OpenAPI 3.1 specification. Integrate detection, forecasting, case management, and configuration into your own workflows, BI tools, and orchestration platforms without touching the UI.

OpenAPI 3.1 Specification

Interactive Swagger UI and ReDoc documentation with request/response schemas and examples for every endpoint.

Dual Authentication

JWT Bearer tokens for SaaS users. X-API-Key header for API-only programmatic access. Both support org-scoped access control.

Role-Based Access Control

Admin, Analyst, and Viewer roles enforced at the API layer — not just the UI. API keys carry scoped permissions.

Fully Multi-Tenant

Every endpoint is scoped to an org slug. One platform, many clients — with strict data isolation at the API and database layers.

REST API · OpenAPI 3.1

# Authenticate with API key

curl -H "X-API-Key: tiq_live_••••••••"

-H "Content-Type: application/json"

https://api.transiq.app/orgs/acme-bank/

results/anomalies

# Run batch inference

POST /orgs/{slug}/results/batch-infer

# Generate AI explanation

POST /orgs/{slug}/explain/anomaly/{id}

# Manage org members & API keys

POST /orgs/{slug}/members

POST /orgs/{slug}/api-keys

# Full OpenAPI docs at:

GET /api/docs → Swagger UI

GET /api/redoc → ReDoc

Built for Your Industry

The same intelligence platform, configured for the specific domains, field schemas, and risk patterns of your sector.