The Halo platform
One control plane for simple, secure, efficient AI
Halo brings discovery, runtime security, governance, and cost control together, so the AI your teams use and the agents you deploy follow the same rules and leave the same evidence.
Four pillars
Discover, secure, govern, and optimize
Each pillar stands on its own and gets stronger with the others, because they share identity, policy, and evidence.
Discover
See the AI your teams use
Find AI apps, extensions, developer tools, and local runtimes across supported browsers and devices, then keep a living inventory.
Secure
Protect governed actions at runtime
Run agents in a sandboxed runtime and route model and tool traffic through gateways that apply identity, policy, and approvals.
Govern
Prove what happened and why
Write policy as code, roll it out in observe or enforce mode, and keep exportable evidence for reviewers.
Optimize
Spend with intent
Set budgets that act on their own, match workloads to the right approved model, and launch agents that do useful work.
Where Halo runs
Enforcement where the work happens
AI shows up in browsers, on laptops, in model calls, and inside agents. Halo meets it in each place.
Browsers
Halo Browser for Chrome and Edge, plus Safari on macOS sees AI sites and extensions and applies audit, warn, or block rules.
Devices
Halo Endpoint for macOS and Windows finds AI apps, developer tools, local runtimes, and add-ins.
Gateways
Model, MCP and tool, and API gateways apply identity, policy, approvals, budgets, and evidence to traffic.
Agent runtimes
Agents run sandboxed with default-deny egress, brokered credentials, and quarantine controls.
FAQ
Questions about the platform
No. Most teams start with one motion, such as discovering AI use on browsers and devices or launching a first governed agent, then add gateways, policy, and budgets as the highest-value use cases become clear.
The same policy model runs at several enforcement points. Browser rules apply in Halo Browser, gateway policy applies to model and tool traffic, and runtime boundaries apply inside the agent sandbox. Decisions from each place land in one evidence record.
Secure agents, the model and tool gateways, policy, approvals, budgets, and evidence are available now. Browser and endpoint discovery roll out with guided onboarding, and the AI gateway for your own applications is in preview.
Halo adds AI-specific visibility and control where general tools have limited context, such as agent tool calls, model requests, and AI browser activity. It works alongside identity, endpoint, and data protection tools you already run.
Put a halo around the AI your business runs.
See discovery, runtime policy, approvals, and evidence working together on your own use cases.