CMDB AI

Tovana Intelligent Automations

Tovana is Sandeza’s AI-driven CMDB accuracy engine, purpose-built to build and maintain a consistent, accurate CMDB for Enterprises. Traditional methods to address CMDB quality fail over time either due to dependence on manual processes, flat-table methods OR changes in validation processes. By leveraging our proprietary, graph-based ensemble model, the CMDB is treated as a relationship graph, and Agentic AI-driven automated remediation, Tovana delivers a feature-rich, self-correcting CMDB that vastly improves enterprise ITSM functionality, and reduces MTTR.

What Tovana Does

 

Data Quality

Automatically evaluates every CI for accuracy, completeness, freshness, and relationship integrity. Multi-factor scoring replaces simple date-based checks to surface CIs that are truly stale or operationally inactive.

Graph-Based Detection

Leverages a graph-based ML model that maps CI inter-relationships. The Analyses connectivity patterns, relationship depth, and structural position to detect orphaned devices, missing owners, duplicate CIs, and invalid relationships.

CMDB Auto-Heal

Continuously enriches and corrects CMDB records using ML-driven confidence scoring. High-confidence anomalies are auto-remediated; low-confidence ones are queued for CMDB manager review- and reinforced learning for future.

Tovana-Features at a glance

Get a complete real-time overview of your CMDB health at a glance. Track key metrics — CI nodes, edges, service roots, anomalies, anomaly rate, and components — in one unified view. Visualise risk score distribution across bands (0–19, 20–39, 40–59, 60–79), reason breakdowns (Missing Host, Under-Connected, Missing Upstream, Missing Workload), and CI class groups such as Application, Server, Network, and Storage.


Every Configuration Item is ranked by risk score and surfaced in a prioritised list. Filter by CI class, anomaly reason, or risk level to instantly focus on the records that matter most. Each row shows the CI System ID, class, group, colour-coded risk score bar, and the reason flag (e.g. Missing Host) — so your CMDB manager always knows exactly where to act first. Recent scan results are shown alongside, keeping the team informed of the latest model run.


Drill into any CI to see its full identity, graph position, and relationship flags side by side. The Ensemble Model Scores panel exposes individual algorithm outputs — Isolation Forest, LOF, One-Class SVM, DBSCAN — alongside the combined Ensemble score and a Model Agreement count. Relationship flags (Compute, Containment, Network, Dependency, Cluster, Storage) pinpoint exactly which connections are missing or broken, with step-by-step remediation guidance for every anomaly reason.


Zoom out to see your entire CMDB landscape — 142+ CIs and 238+ relationships rendered as a live topology map. Instantly spot isolated clusters, orphaned CIs, and data quality hotspots that traditional list views would miss. CIs floating alone or in small disconnected clusters are immediate signals of relationship gaps or stale records that Tovana automatically flags for remediation.


Explore every CI and relationship in your CMDB as an interactive force-directed graph. Green nodes indicate healthy CIs; orange nodes flag anomalies needing attention. Click any node to instantly see its risk score, connectivity stats (In/Out degree, Component size, Hops to root), and a recommended “What to Do” action list — all without leaving the graph view.