Govern · Governance Dashboard
Available nowSee AI activity, decisions, and cost together
Halo dashboards show audit statistics, decision timeseries, AI inventory views, and cost analytics, so security, governance, and finance teams can ask questions of the same data.
At a glance
- views, audit, decisions, inventory, and cost
- 4
Why it matters
AI oversight is split across too many screens
Inventory lives in one place, spend in another, and policy decisions in a third. Leaders end up asking for a custom report each time a question comes up.
- 01
Security reviewers need to watch decisions and denials over time to spot a change in behavior.
- 02
Governance leads need an inventory view that shows what is sanctioned and what is still unreviewed.
- 03
Finance and AI operations owners need cost by agent and model before invoices arrive.
Cost and governance
One place for spend and decisions
Cost analytics sit next to audit statistics and inventory, so a spike in spend can be traced to an agent, a model, and the policy decisions around the same time. Finance and security can look at the same facts.
- Cost analytics by agent and model
- Audit statistics over selectable windows
- Inventory views by classification and surface
- Attention panels for unhealthy or quarantined agents
Capabilities
Four views on the same records
Each view reads from the evidence Halo already collects, so numbers can be traced to the records behind them.
- Audit statistics
- Audit statistics summarize events by type over a window you select, so an unusual jump in policy or configuration activity stands out.
- Decision timeseries
- Charts show policy decisions over time, so reviewers can see how outcomes shift as usage grows or a rule changes.
- Inventory views
- Inventory views group AI items by classification, surface, and risk level, with the unreviewed queue easy to find.
- Cost analytics
- Cost analytics break spend down by agent and model across time windows, so owners can see where usage and cost concentrate.
- Agents needing attention
- Fleet views flag agents that are unhealthy, quarantined, or waiting on approvals, so operators know where to look first.
- Drill down to evidence
- Move from a chart to the records behind it, including run timelines and evidence checkpoints, and export what you find.
How it works
From activity to answers
The dashboards read from the same records that gateways, the runtime, and endpoints create.
- 1
Activity is recorded
Browsers, endpoints, gateways, and the agent runtime report what they observe and decide, and the control plane stores the records.
- 2
Choose a view
Open audit statistics, decisions, inventory, or cost, and set the time window that fits the question you are asking.
- 3
Investigate and share
Drill into the records behind a number, then export the evidence for a review, a ticket, or a finance conversation.
FAQ
Questions, answered
Halo dashboards cover audit statistics, decision timeseries, inventory views, and cost analytics. Fleet views also surface agents that need attention, such as unhealthy or quarantined agents and pending approvals, with time windows you can change.
Security leaders use decision and audit views, governance and compliance leads use inventory and evidence views, and finance and AI operations owners use cost analytics. Each group works from the same underlying records.
Halo does not produce a compliance score or map controls to a law automatically. The dashboards show evidence and activity that support governance programs aligned to frameworks such as NIST AI RMF, and your team interprets what they show.
Audit logs and gateway decisions export as CSV and JSON, and the AI inventory exports as an AI BOM in CycloneDX, JSON, or CSV. Apply filters first to scope an export to the period or items under review.
Access follows account roles. Members can view agents, monitoring, and cost dashboards, and account admins also see audit logs and manage settings. Your identity provider can control who signs in through OIDC single sign-on.
Related capabilities
Put a halo around the AI your business runs.
See discovery, runtime policy, approvals, and evidence working together on your own use cases.