# Halo by OmniTrust

> Halo is an AI governance platform. It discovers the AI in use on supported browsers and devices, secures agents and model and tool traffic with gateways and a sandboxed runtime, governs with policy as code and exportable evidence, and keeps AI spend efficient with budgets and model routing.

The full text of every page is in one Markdown file at https://omnitrusthalo.com/llms-full.txt.

## Company and product

- [Halo | AI governance platform for secure, efficient AI](https://omnitrusthalo.com/): Halo discovers the AI in use, enforces policy in browsers, on devices, at model and tool gateways, and in agent runtimes, and keeps evidence reviewers trust.
- [AI Governance Platform | Halo by OmniTrust](https://omnitrusthalo.com/ai-governance-platform): Halo finds the AI in use across your organization, enforces policy live on devices, browsers, gateways, and agents, and keeps the evidence your reviewers need.
- [Halo pricing | Plans for secure AI agents and governance](https://omnitrusthalo.com/pricing): Start with one secure agent for $99 per month or talk with us about enterprise governance. Plans include gateways, budget controls, and audit evidence.
- [Resource center | AI governance guides and checklists | Halo](https://omnitrusthalo.com/resources): Guides, checklists, comparisons, and definitions to help your team discover AI use, govern agents, and build evidence reviewers trust.
- [Halo FAQ | Discovery, agents, governance, and pricing](https://omnitrusthalo.com/faq): Answers to common questions about Halo discovery, secure agents, gateways, policy, evidence, privacy, plans, and getting started.
- [About Us | The company behind Halo | OmniTrust](https://omnitrusthalo.com/about): Explore OmniTrust security heritage, expertise, and patent portfolio, and how that experience shapes Halo and the way we work with customers.
- [Partner with Halo | Technology, referral, and delivery](https://omnitrusthalo.com/partners): Work with OmniTrust to deliver governed AI. Explore technology, referral, and implementation partnerships with Halo.
- [Security & Privacy at Halo | OmniTrust](https://omnitrusthalo.com/security): How Halo protects customer data and agents, from brokered credentials and sandboxed runtimes to metadata-first discovery and exportable evidence.
- [Contact Halo | Talk with our team](https://omnitrusthalo.com/contact-sales): Talk with the Halo team about secure agents, AI visibility, governance, pricing, and services.
- [Halo Support | OmniTrust](https://omnitrusthalo.com/support): Get help with Halo products, deployments, enrollment, and account access.
- [Terms of Use | Halo by OmniTrust](https://omnitrusthalo.com/legal/terms): The terms that govern use of Halo services and websites provided by OmniTrust.
- [Privacy Policy | Halo by OmniTrust](https://omnitrusthalo.com/legal/privacy): How OmniTrust collects, uses, and protects personal information for Halo.
- [Browser Privacy Notice | Halo by OmniTrust](https://omnitrusthalo.com/legal/browser-privacy): What the Halo Browser extension collects, what it does not collect, and the settings your organization controls.
- [Endpoint Privacy Notice | Halo by OmniTrust](https://omnitrusthalo.com/legal/endpoint-privacy): What Halo Endpoint collects on enrolled devices, what it does not collect, and how the data is retained.
- [Acceptable Use Policy | Halo by OmniTrust](https://omnitrusthalo.com/legal/aup): The uses of Halo services, agents, and Halo Chat that are not permitted.

## Platform

- [The Halo platform | Discover, secure, govern, optimize](https://omnitrusthalo.com/platform): One control plane to discover AI use, secure agents and model traffic, govern with policy and evidence, and keep AI spend efficient.
- [AI Discovery Across Browsers and Endpoints | Halo](https://omnitrusthalo.com/platform/ai-discovery): Halo AI Discovery finds the AI sites, browser extensions, apps, and local runtimes your teams use on supported surfaces, with source and confidence.
- [AI Inventory and AI BOM Export in CycloneDX | Halo](https://omnitrusthalo.com/platform/ai-inventory): Build an AI inventory with Halo. Classify apps as sanctioned, monitored, blocked, shadow, or unknown, and export an AI BOM in CycloneDX, JSON, or CSV.
- [AI Browser Extension for Chrome, Edge, and Safari | Halo](https://omnitrusthalo.com/platform/halo-browser): Halo Browser is an AI browser extension for Chrome and Edge, plus Safari on macOS that discovers AI site use and extensions, with audit, warn, and block controls.
- [Endpoint AI Discovery for macOS and Windows | Halo](https://omnitrusthalo.com/platform/halo-endpoint): Endpoint AI discovery with Halo Endpoint surveys AI apps, IDE plug-ins, local runtimes, and model locations on macOS and Windows, metadata first.
- [Secure AI Agent Runtime with Default-Deny Egress | Halo](https://omnitrusthalo.com/platform/agent-runtime): Run AI agents in a secure agent runtime with sandboxed, non-root workloads, default-deny network egress, signed configuration, and quarantine controls.
- [MCP Gateway for Governed AI Tool Access | Halo](https://omnitrusthalo.com/platform/mcp-gateway): The Halo MCP gateway puts identity, policy, approvals, budgets, and evidence in front of agent tool calls and brokers credentials so agents never see raw keys.
- [AI Model Gateway with BYOK and Managed Models | Halo](https://omnitrusthalo.com/platform/model-gateway): The Halo model gateway applies policy, budgets, and evidence to model requests, with Halo-managed models or your own OpenAI, Anthropic, or Gemini key.
- [Runtime Guardrails for AI Prompts and Responses | Halo](https://omnitrusthalo.com/platform/runtime-guardrails): Halo runtime guardrails run selected detectors for PII, secrets, prompt injection, moderation, and language in supported gateway paths, with masking.
- [Human Approval Workflows for AI Agent Actions | Halo](https://omnitrusthalo.com/platform/agent-approvals): Halo human approvals pause higher-impact AI agent actions until a reviewer decides, then record the request, the decision, and the note as evidence.
- [AI Policy as Code with Audit and Enforce Modes | Halo](https://omnitrusthalo.com/platform/policy-engine): Write AI policy as code with Halo. Versioned policy bundles return allow, deny, or alert outcomes, run in audit or enforce mode, and explain each decision.
- [AI Audit Trail and Evidence Export | Halo](https://omnitrusthalo.com/platform/audit-evidence): Halo keeps an AI audit trail of decisions, approvals, denials, costs, and configuration changes, and exports audit logs and gateway decisions as CSV and JSON.
- [AI Governance Dashboard for Security and Finance | Halo](https://omnitrusthalo.com/platform/governance-dashboard): The Halo AI governance dashboard shows audit statistics, decision timeseries, inventory views, and cost analytics for security, governance, and finance teams.
- [AI Agent Identity and Access Management | Halo](https://omnitrusthalo.com/platform/agent-identity): AI agent identity in Halo is separate from user identity, credentials, tools, and action scope, so access can be scoped, attributed, and contained.
- [AI Model Routing with Route Profiles and Fallback | Halo](https://omnitrusthalo.com/platform/model-routing): AI model routing in Halo uses route profiles, model catalogs with pricing, and fallback paths in the model gateway to match workloads to approved models.
- [AI Budget Controls and Cost Analytics for Agents | Halo](https://omnitrusthalo.com/platform/cost-controls): Set AI budget controls in Halo with account and per-agent budgets, thresholds, and automatic actions that notify, pause new operations, or stop deployments.
- [Launch Secure AI Agents with Halo Agents | Halo](https://omnitrusthalo.com/platform/halo-agents): Halo Agents launches secure AI agents with model choice, app connections, policy, approvals, and budget set together. Self-serve plans start at $99 per month.
- [AI App Connections with 1,000+ Apps in the Catalog | Halo](https://omnitrusthalo.com/platform/app-connections): AI app connections in Halo link agents to the connection catalog of 1,000+ apps, with scoped access, brokered credentials, policy, and approvals.

## Solutions

- [AI governance solutions by problem and team | Halo](https://omnitrusthalo.com/solutions): Start with shadow AI, agent governance, policy enforcement, AI spend, or compliance evidence, and see which teams and Halo capabilities each one involves.
- [Shadow AI Discovery for Browsers and Devices | Halo](https://omnitrusthalo.com/solutions/shadow-ai-discovery): Shadow AI discovery combines guided browser and endpoint signals with classified inventory, source and confidence, and exportable AI BOM records.
- [AI Agent Governance for Model and Tool Access | Halo](https://omnitrusthalo.com/solutions/ai-agent-governance): AI agent governance connects distinct identities to scoped tool and model access, approvals, runtime controls, and evidence for review.
- [AI Policy Enforcement for Browsers and Agents | Halo](https://omnitrusthalo.com/solutions/ai-policy-enforcement): AI policy enforcement takes structured rules from audit mode to enforcement across supported browsers, model and tool gateways, and agent paths.
- [AI Cost Governance by Agent and Model | Halo](https://omnitrusthalo.com/solutions/ai-cost-governance): AI cost governance connects agent and model analytics with account and per-agent budgets and configurable routing with savings potential.
- [AI Compliance Evidence for Governance Teams | Halo](https://omnitrusthalo.com/solutions/ai-compliance-evidence): AI compliance evidence connects inventory, decisions, approvals, and exports to support governance programs and internal review.

## Integrations

- [Integrations | Models, browsers, devices, and apps | Halo](https://omnitrusthalo.com/integrations): Connect Halo to model providers, browsers, devices, MCP servers, and 1,000+ apps in the connection catalog, with policy and evidence on each connection.
- [OpenAI integration for governed AI | Halo](https://omnitrusthalo.com/integrations/openai): Route OpenAI model access through Halo with brokered credentials, policy checks, approval gates, budget controls, and reviewable evidence for agent traffic.
- [Anthropic integration for governed AI | Halo](https://omnitrusthalo.com/integrations/anthropic): Connect Anthropic model access through Halo with BYOK credential brokering, policy checks, budget controls, approvals, and reviewable agent evidence.
- [Google Gemini BYOK integration for Halo | Halo](https://omnitrusthalo.com/integrations/google-gemini): Bring your own Gemini API key to Halo and route Gemini model requests through the governed model gateway with policy, budgets, approvals, and evidence.
- [Amazon Bedrock integration and Halo model access | Halo](https://omnitrusthalo.com/integrations/amazon-bedrock): Halo-managed models run through Amazon Bedrock, so agents get governed model access with policy, budgets, and evidence and no provider key to manage.
- [Google Chrome integration for AI governance | Halo](https://omnitrusthalo.com/integrations/google-chrome): Use Halo Browser with Chrome for guided discovery of supported AI activity and browser policy controls with audit, warn, and block modes.
- [Microsoft Edge integration for AI governance | Halo](https://omnitrusthalo.com/integrations/microsoft-edge): Roll out Halo Browser in Microsoft Edge for guided discovery of supported AI activity and audit, warn, or block browser policy controls.
- [Safari integration for Halo browser governance | Halo](https://omnitrusthalo.com/integrations/safari): Bring Halo Browser to Safari on macOS through guided onboarding, with discovery and policy controls scoped to supported AI sites.
- [macOS integration for Halo AI discovery | Halo](https://omnitrusthalo.com/integrations/macos): Deploy Halo Endpoint on macOS for metadata-first discovery of AI applications, runtimes, artifacts, and related configuration signals.
- [Windows integration for Halo AI discovery | Halo](https://omnitrusthalo.com/integrations/windows): Roll out Halo Endpoint on Windows for metadata-first discovery of AI applications, runtimes, artifacts, and related configuration signals.
- [Model Context Protocol integration and governance | Halo](https://omnitrusthalo.com/integrations/model-context-protocol): Route Model Context Protocol traffic through Halo’s MCP and tool gateway to apply identity, policy, approval, budget, and evidence controls.
- [CycloneDX integration for Halo AI BOM export | Halo](https://omnitrusthalo.com/integrations/cyclonedx): Export Halo AI inventory as an AI bill of materials in CycloneDX, alongside JSON and CSV, for review in your existing governance workflows.
- [Slack integration for governed AI agents | Halo](https://omnitrusthalo.com/integrations/slack): Connect Slack through the Halo connection catalog and govern agent access with scoped actions, credential brokering, approvals, budgets, and evidence.
- [Microsoft 365 integration for Halo AI agents | Halo](https://omnitrusthalo.com/integrations/microsoft-365): Connect Microsoft 365 through Microsoft Graph in the Halo catalog, with scoped agent actions, credential brokering, approvals, budgets, and evidence.
- [GitHub integration for governed AI agents | Halo](https://omnitrusthalo.com/integrations/github): Connect GitHub through Halo’s catalog to govern agent actions with scoped capabilities, brokered credentials, approvals, budgets, and reviewable evidence.
- [GitLab integration for governed AI agents | Halo](https://omnitrusthalo.com/integrations/gitlab): Connect GitLab through Halo’s catalog and govern configured agent actions with scoped access, credential brokering, approvals, budgets, and evidence.
- [Jira integration for governed AI agents | Halo](https://omnitrusthalo.com/integrations/jira): Connect Jira through Halo’s connection catalog and govern configured agent actions with scoped access, approvals, credential brokering, and evidence.
- [Linear integration for governed AI agents | Halo](https://omnitrusthalo.com/integrations/linear): Connect Linear through Halo’s app catalog and govern configured agent actions with scoped access, credential brokering, approvals, budgets, and evidence.
- [Asana integration for governed AI agents | Halo](https://omnitrusthalo.com/integrations/asana): Connect Asana through Halo’s connection catalog and govern configured agent actions with scoped capabilities, brokered credentials, approvals, and evidence.
- [Notion integration for governed AI agents | Halo](https://omnitrusthalo.com/integrations/notion): Connect Notion through Halo’s connection catalog and govern agent access with scoped actions, brokered credentials, approval gates, and reviewable evidence.
- [Airtable integration for governed AI agents | Halo](https://omnitrusthalo.com/integrations/airtable): Connect Airtable through Halo’s connection catalog and govern configured agent actions with scoped access, credential brokering, approvals, and evidence.
- [Intercom integration for governed AI agents | Halo](https://omnitrusthalo.com/integrations/intercom): Connect Intercom through Halo’s connection catalog and govern agent access to supported conversations and contacts with scoped actions, approvals, and evidence.
- [HubSpot integration for governed AI agents | Halo](https://omnitrusthalo.com/integrations/hubspot): Connect HubSpot through Halo’s connection catalog and govern configured agent actions with scoped access, credential brokering, approval gates, and evidence.
- [Zendesk integration for governed AI agents | Halo](https://omnitrusthalo.com/integrations/zendesk): Connect Zendesk through Halo’s catalog and govern configured agent access with scoped actions, credential brokering, approvals, budgets, and evidence.
- [Snowflake integration for governed AI agents | Halo](https://omnitrusthalo.com/integrations/snowflake): Connect Snowflake through Halo’s connection catalog and govern configured agent access with scoped actions, brokered credentials, approvals, and evidence.
- [Datadog integration for governed AI agents | Halo](https://omnitrusthalo.com/integrations/datadog): Connect Datadog through Halo’s connection catalog and govern configured agent actions with scoped access, credential brokering, approvals, and evidence.
- [Sentry integration for governed AI agents | Halo](https://omnitrusthalo.com/integrations/sentry): Connect Sentry through Halo’s catalog to govern configured agent access with scoped actions, credential brokering, approvals, budgets, and evidence.
- [PagerDuty integration for governed AI agents | Halo](https://omnitrusthalo.com/integrations/pagerduty): Connect PagerDuty through Halo’s connection catalog and govern configured agent access with scoped actions, approvals, credential brokering, and evidence.
- [Stripe integration for governed AI agents | Halo](https://omnitrusthalo.com/integrations/stripe): Connect Stripe through Halo’s connection catalog and govern configured agent access with scoped actions, brokered credentials, approvals, and evidence.

## Comparisons

- [Compare Halo with other AI governance approaches](https://omnitrusthalo.com/compare): See how Halo compares with AI governance, AI security, and AI gateway vendors and with DIY governance, in sourced side-by-side comparisons.
- [Halo vs Airia | AI governance platform comparison](https://omnitrusthalo.com/compare/halo-vs-airia): Compare Halo and Airia. Halo enforces AI policy on devices, in browsers, at gateways, and in a sandboxed agent runtime, with budgets that act and published pricing.
- [Halo vs Geordie AI | AI governance platform comparison](https://omnitrusthalo.com/compare/halo-vs-geordie): Compare Halo and Geordie AI. Halo governs AI use by people and agents in browsers, on endpoints, at gateways, and in a sandboxed runtime, with budgets that act.
- [Halo vs Onyx Security | AI governance platform comparison](https://omnitrusthalo.com/compare/halo-vs-onyx-security): Compare Halo and Onyx Security. Halo runs its own browser and endpoint sensors, enforces at gateways and in a sandboxed agent runtime, and keeps policy as reviewable code.
- [Halo vs OneTrust | AI governance platform comparison](https://omnitrusthalo.com/compare/halo-vs-onetrust): Compare Halo and OneTrust. Halo is built for AI, with browser and endpoint discovery and live enforcement at model and tool gateways and in a sandboxed agent runtime.
- [Halo vs Credo AI | AI governance platform comparison](https://omnitrusthalo.com/compare/halo-vs-credo-ai): Compare Halo and Credo AI. Halo enforces AI policy in the request path today, with browser and endpoint discovery, gateways, approvals, budgets, and a sandboxed runtime.
- [Halo vs ServiceNow | AI governance platform comparison](https://omnitrusthalo.com/compare/halo-vs-servicenow): Compare Halo and ServiceNow AI Control Tower. Halo is built for AI, needs no other platform, and enforces policy on devices, in browsers, at gateways, and on agents.
- [Halo vs IBM watsonx.governance | AI governance comparison](https://omnitrusthalo.com/compare/halo-vs-ibm-watsonx-governance): Compare Halo and IBM watsonx.governance. Halo finds the AI people and agents use, including tools nobody registered, and enforces policy on it as the work happens.
- [Halo vs Holistic AI | AI governance platform comparison](https://omnitrusthalo.com/compare/halo-vs-holistic-ai): Compare Halo and Holistic AI. Halo enforces AI policy as the work happens, with browser and endpoint discovery, gateways, a sandboxed agent runtime, and budgets that act.
- [Halo vs AI gateways | AI governance platform comparison](https://omnitrusthalo.com/compare/halo-vs-ai-gateways): Compare Halo with AI gateways. Halo includes model and tool gateways and adds discovery, a sandboxed agent runtime, budgets, and evidence for AI beyond the gateway.
- [Halo vs DIY governance | AI governance comparison](https://omnitrusthalo.com/compare/halo-vs-diy-governance): Compare Halo with DIY AI governance built from spreadsheets, scripts, and point tools. See how Halo replaces manual upkeep with discovery, enforcement, and exportable evidence.

## Blog

- [Halo blog | AI governance, agent security, and AI efficiency](https://omnitrusthalo.com/blog): Weekly field guides on AI governance, shadow AI, MCP security, agent identity, compliance evidence, and keeping AI spend efficient.
- [Build AI budgets that follow the owner | Halo blog](https://omnitrusthalo.com/blog/ai-budgets-that-follow-the-owner): AI budgets work when each agent has an owner, thresholds trigger clear actions, and exceptions leave a record. See how account and per-agent budgets fit together.
- [Who reviews the governance assistant? | Halo blog](https://omnitrusthalo.com/blog/reviewing-governance-assistant): Assistants that draft governance work need provenance, permission boundaries, and an accountable human decision. Learn how to review their suggestions before anything is approved.
- [What ISO/IEC 42001 asks of your AI software | Halo blog](https://omnitrusthalo.com/blog/iso-42001-and-your-ai-software): ISO/IEC 42001 sets requirements for an organization's AI management system. Learn what auditors may ask of your AI tools and what evidence software can supply for that system.
- [How to evaluate an AI governance platform in one workflow | Halo](https://omnitrusthalo.com/blog/evaluate-ai-governance-platform-workflow): Use one realistic workflow to test discovery, policy, approval, failure behavior, and evidence. This run sheet works for any AI governance platform, including Halo.
- [How to respond to an AI agent incident | Halo blog](https://omnitrusthalo.com/blog/ai-incident-response-for-agents): When an agent misbehaves, the steps are contain, preserve evidence, revoke credentials, review, and recover. Here is a plan you can rehearse before you need it.
- [When should an AI system be reviewed again? | Halo blog](https://omnitrusthalo.com/blog/when-to-review-ai-system-again): A review records a decision under particular conditions. Learn which changes should reopen it, how to scope the work, and how to keep both decisions on the record.
- [How to roll out AI controls without surprising your teams | Halo](https://omnitrusthalo.com/blog/rolling-out-ai-controls-without-surprises): Start in observe mode, move to warn, then enforce. Measure false positives, communicate each change, and keep a path back so the teams you govern keep trusting the controls.
- [AI cost estimates need a coverage label | Halo blog](https://omnitrusthalo.com/blog/ai-cost-estimate-coverage): Before an AI spend number guides investment, label what it covers, how it was priced, and what it leaves out. Learn how to reconcile estimates with provider bills.
- [NIST AI RMF, from framework to runtime evidence | Halo blog](https://omnitrusthalo.com/blog/nist-ai-rmf-runtime-evidence): The NIST AI RMF organizes work into Govern, Map, Measure, and Manage. See how each function connects to operational controls and to runtime evidence a reviewer can inspect.
- [What happens when an AI control goes offline? | Halo blog](https://omnitrusthalo.com/blog/ai-control-outage-behavior): AI controls depend on other systems. Map the dependencies, choose failure behavior by action, and test outages with synthetic data before production decides for you.
- [What is an AI bill of materials, and who uses it? | Halo blog](https://omnitrusthalo.com/blog/ai-bill-of-materials): An AI bill of materials lists the models, tools, data sources, and services behind your AI. See what it holds, how CycloneDX fits, and how security, procurement, and audit use it.
- [A practical checklist for MCP governance | Halo blog](https://omnitrusthalo.com/blog/mcp-governance-checklist): Follow one agent tool call from server inventory to authorization, approval, execution, and evidence with this seven-step MCP governance checklist for security teams.
- [Smart model routing that keeps governance intact | Halo blog](https://omnitrusthalo.com/blog/smart-model-routing-governance): Cost-aware routing can match each workload to an approved model that fits its cost, quality, and policy needs. Here is how to pursue that savings potential and keep control.
- [An agent's permission is only the start of the decision | Halo](https://omnitrusthalo.com/blog/agent-permission-and-action-decisions): A valid credential shows what an agent may request. Learn how purpose, scope, arguments, and approval shape whether one specific action should run.
- [Browser AI needs its own governance model | Halo blog](https://omnitrusthalo.com/blog/browser-ai-governance): Browser AI spans websites, extensions, uploads, and pastes. A workable governance model uses audit, warn, and block modes, domain rules, and clear privacy boundaries.
- [Connecting AI spend to AI value | Halo blog](https://omnitrusthalo.com/blog/ai-spend-versus-ai-value): A token-cost dashboard shows where the money goes. Learn how to connect AI spend to owners, outcomes, cost boundaries, and honest assumptions before you scale.
- [The EU AI Act timeline, translated for governance teams | Halo](https://omnitrusthalo.com/blog/eu-ai-act-timeline-for-governance-teams): The EU AI Act phases in over several years. See what each date means as a decision your team makes and a record you keep, and confirm the current official timeline first.
- [What should an AI audit log contain? | Halo blog](https://omnitrusthalo.com/blog/useful-ai-audit-log): A useful AI audit log lets a reviewer reconstruct a consequential decision. Learn which fields to record, how to handle prompt content, and how to test the result.
- [Give AI agents identities you can revoke | Halo blog](https://omnitrusthalo.com/blog/ai-agent-identity): Agents that borrow a person's login or a shared service account leave reviewers guessing. Give each agent its own identity, scoped credentials, and a path to revocation.
- [A policy is only useful if you can show where it runs | Halo](https://omnitrusthalo.com/blog/where-ai-policy-runs): Turn an AI policy into a testable rule, name the point where it runs, and keep the evidence. This guide shows how to check each step before you enforce anything.
- [MCP security from server inventory to runtime enforcement | Halo](https://omnitrusthalo.com/blog/mcp-security-from-inventory-to-runtime): MCP servers behave like a supply chain. Learn how to inventory them, authorize them, scope their credentials, apply tool-call policy, and keep evidence reviewers can use.
- [Your AI inventory is a moving target | Halo blog](https://omnitrusthalo.com/blog/ai-inventory-coverage-and-freshness): An AI inventory is only as useful as its coverage and freshness. Learn how to record where you looked, when you last looked, and which changes need a new decision.
- [Why shadow AI discovery has to reach browsers and endpoints | Halo](https://omnitrusthalo.com/blog/shadow-ai-discovery-browsers-and-endpoints): Gateway and network logs miss local apps, browser extensions, IDE tools, and local runtimes. See how metadata-first discovery on browsers and endpoints closes the gap.
- [What is an AI governance platform? | Halo blog](https://omnitrusthalo.com/blog/what-is-an-ai-governance-platform): An AI governance platform connects AI inventory, decisions, runtime controls, and evidence for each system. Learn what it does and how to test one with a real use case.

## Glossary

- [AI governance glossary | Terms defined plainly | Halo](https://omnitrusthalo.com/glossary): Plain definitions of AI governance, AI security, and agent terms, from shadow AI and MCP gateways to policy as code and the EU AI Act.
- [What is agent identity? AI access explained | Halo](https://omnitrusthalo.com/glossary/agent-identity): Agent identity distinguishes an AI agent from its user and helps define its allowed tools, credentials, and actions. Learn how Halo separates these scopes.
- [Agentic AI explained for governance teams | Halo](https://omnitrusthalo.com/glossary/agentic-ai): Agentic AI describes systems that use models to plan or take actions through tools. Learn how identity, permissions, and approvals shape governance.
- [What is an AI agent? Definition and controls | Halo](https://omnitrusthalo.com/glossary/ai-agent): An AI agent uses a model to carry out a task and may call tools. Learn how Halo applies identity, policy, approval, budget, and evidence controls.
- [AI audit trail meaning and governance context | Halo](https://omnitrusthalo.com/glossary/ai-audit-trail): An AI audit trail records decisions, approvals, denials, costs, and configuration changes for review. Learn what Halo makes exportable.
- [What is an AI BOM? Definition and uses | Halo](https://omnitrusthalo.com/glossary/ai-bill-of-materials): An AI BOM is a structured record of AI components in use. Learn how Halo exports inventory data in CycloneDX, JSON, and CSV for review.
- [What is the AI control plane? | Halo](https://omnitrusthalo.com/glossary/ai-control-plane): An AI control plane coordinates identity, policy, approvals, and evidence across supported AI models, tools, agents, and discovery surfaces.
- [AI cost governance explained | Halo](https://omnitrusthalo.com/glossary/ai-cost-governance): AI cost governance combines budgets, thresholds, and cost analytics for models and agents. Learn how Halo supports controls without promising fixed savings.
- [AI data loss prevention explained | Halo](https://omnitrusthalo.com/glossary/ai-data-loss-prevention): AI data loss prevention applies checks to sensitive information in AI traffic. Learn where Halo supports regex checks, masking, and redaction.
- [What is an AI gateway? Definition and controls | Halo](https://omnitrusthalo.com/glossary/ai-gateway): An AI gateway mediates model traffic and applies selected controls. Learn how Halo routes agent requests and where external app ingress is preview.
- [What is an AI governance platform? | Halo](https://omnitrusthalo.com/glossary/ai-governance-platform): An AI governance platform helps teams discover AI use, set policy, govern agent activity, and review evidence across supported environments.
- [What are AI guardrails? Definition and controls | Halo](https://omnitrusthalo.com/glossary/ai-guardrails): AI guardrails are checks and policies that constrain model and agent activity. Learn which runtime classifiers and controls Halo supports.
- [What is an AI inventory? Definition | Halo](https://omnitrusthalo.com/glossary/ai-inventory): Learn how an AI inventory organizes detected AI applications, their source and confidence, and classifications used to prioritize review.
- [AI risk classification explained | Halo](https://omnitrusthalo.com/glossary/ai-risk-classification): AI risk classification gives teams a consistent way to categorize systems and prioritize review. Learn which status labels Halo uses in its inventory.
- [AI runtime security explained for agents | Halo](https://omnitrusthalo.com/glossary/ai-runtime-security): AI runtime security applies controls while agents execute. Learn how Halo combines sandboxing, default-deny egress, identity boundaries, and approvals.
- [Bring your own key explained for AI models | Halo](https://omnitrusthalo.com/glossary/bring-your-own-key): Bring your own key lets an organization use its own model provider credential through a supported gateway. Learn which providers Halo supports for BYOK.
- [What is credential brokering? Definition | Halo](https://omnitrusthalo.com/glossary/credential-brokering): Credential brokering keeps provider or app secrets at a gateway while agents receive scoped capabilities. Learn how Halo handles model and app access.
- [Default-deny egress explained for AI security | Halo](https://omnitrusthalo.com/glossary/default-deny-egress): Default-deny egress blocks agent network traffic unless an explicit boundary allows it. Learn how Halo uses this control in its secure runtime.
- [EU AI Act overview for governance teams | Halo](https://omnitrusthalo.com/glossary/eu-ai-act): Learn the EU AI Act timeline and its phased obligations, including key dates listed in current guidance. Confirm the official timeline and seek legal advice.
- [Human in the loop explained for AI agents | Halo](https://omnitrusthalo.com/glossary/human-in-the-loop): Human in the loop adds a person’s review or approval to an AI workflow. Learn how Halo supports approval gates for higher-impact actions.
- [ISO/IEC 42001 overview for AI governance | Halo](https://omnitrusthalo.com/glossary/iso-iec-42001): ISO/IEC 42001 is an AI management system standard published in 2023. Learn how Halo evidence may support governance work without implying certification.
- [What is an MCP gateway? Definition | Halo](https://omnitrusthalo.com/glossary/mcp-gateway): An MCP gateway mediates agent requests to MCP tools. Learn how identity, policy, approvals, and evidence apply to supported traffic.
- [Model Context Protocol definition and governance | Halo](https://omnitrusthalo.com/glossary/model-context-protocol): Model Context Protocol connects AI applications with tools and data. Learn what MCP is and how Halo governs supported tool traffic.
- [Model routing explained for AI governance | Halo](https://omnitrusthalo.com/glossary/model-routing): Model routing selects a model for a request based on configured routes and fallbacks. Learn how Halo connects routing with policy and budgets.
- [NIST AI RMF explained for governance teams | Halo](https://omnitrusthalo.com/glossary/nist-ai-rmf): NIST AI RMF 1.0 organizes AI risk work into Govern, Map, Measure, and Manage. Learn how Halo evidence can support a program aligned to the framework.
- [Policy as code explained for AI governance | Halo](https://omnitrusthalo.com/glossary/policy-as-code): Policy as code stores rules in structured, versioned bundles. Learn how Halo evaluates tool and model traffic with audit and enforce modes.
- [Prompt injection explained for AI security | Halo](https://omnitrusthalo.com/glossary/prompt-injection): Prompt injection is input that can try to steer an AI system away from intended instructions. Learn how selected Halo runtime checks fit into governance.
- [Shadow AI explained for governance teams | Halo](https://omnitrusthalo.com/glossary/shadow-ai): Learn what shadow AI means, how unreviewed AI use can create visibility gaps, and how inventory and policy help teams manage supported surfaces.