Discover · Halo Endpoint
Guided rollout· macOS and WindowsFind the AI running on your devices
Halo Endpoint surveys macOS and Windows devices for AI applications, developer tools, local runtimes, model artifact locations, and configuration files, and records paths, hashes, and permissions as evidence. Standard collection stays metadata first.
- Linux coming soon
- Windows and macOS
- standard collection
- Metadata first
Why it matters
Local AI can run without any browser traffic
Developers install AI assistants, run local model runtimes, and download model files that browser monitoring cannot see. Without a sensor on the device, that AI stays invisible to the teams who govern it.
- 01
A local runtime can serve a model on a laptop without any request to an AI provider.
- 02
IDE plug-ins and developer AI tools can send code to external services straight from the editor.
- 03
Startup items and add-ins can bring AI features into everyday applications without a review.
Capabilities
What Halo Endpoint surveys on a device
Collectors cover applications, browsers, IDE plug-ins, runtimes, startup items, add-ins, and configuration files, then match what they find to the Halo AI catalog.
AI applications
Surveys installed AI applications, including AI features delivered through add-ins, and matches findings to the Halo AI catalog on macOS and Windows.
IDE plug-ins and developer tools
Detects IDE plug-ins and developer AI tools, such as coding assistants, so reviewers get a record without asking engineers to fill out a survey.
Local AI runtimes
Spots local runtimes that host models on a device, for example Ollama-style runtimes, which browser monitoring cannot see.
Model artifact locations
Records where model files are stored, with paths, hashes, and permissions as the evidence, so reviewers can find local models.
Startup items and add-ins
Notes startup items and add-ins that bring AI features into everyday tools, so reviewers can see what loads when a device starts.
AI configuration files
Records AI-related configuration files, including MCP configuration in supported locations, as paths, hashes, and permissions, so reviewers can see how tools on a device are set up.
How it works
From package to reviewed findings
Rollout is guided, so your team and Halo agree on scope before the first device reports.
- 1
Package the sensor
Halo's team helps package and distribute the sensor for managed macOS and Windows devices and plan which groups enroll first.
- 2
Enroll each device
Devices enroll with an account-scoped token, begin periodic surveys, and report health, so you can see which devices are reporting.
- 3
Survey and match
The sensor surveys applications, plug-ins, runtimes, startup items, add-ins, and configuration files, then matches what it finds to the Halo AI catalog.
- 4
Review in the inventory
Findings join browser discoveries in the AI inventory with their evidence, ready for classification and export as an AI BOM.
Metadata first
Collection built around metadata
Halo Endpoint captures names, paths, hashes, permissions, and model artifact locations as evidence. Standard collection does not include prompts, responses, documents, clipboard content, or credentials, which keeps reviews focused on what is installed and where it lives.
- Names, paths, hashes, and permissions as evidence
- Prompts and documents excluded from standard collection
- Clipboard content and credentials excluded as well
- The Endpoint Privacy Notice details what is collected
FAQ
Questions, answered
Halo Endpoint supports macOS and Windows. Other operating systems are not supported today. Rollout is guided, so Halo's team helps you package and distribute the sensor to managed devices and confirm which groups to enroll first.
Standard collection records metadata about AI applications, browsers, IDE plug-ins, developer AI tools, local runtimes, model artifact locations, startup items, add-ins, and AI-related configuration files. Evidence includes paths, hashes, and permissions. It does not include prompts, responses, documents, clipboard content, or credentials.
Halo Endpoint reports what it can observe on enrolled devices, in locations and formats it recognizes. A model stored in an unusual place, a tool Halo does not recognize, or a device that is not enrolled may not appear. The guided rollout includes a review of coverage gaps.
Halo's team helps you choose which devices enroll first, package the sensor for macOS and Windows, and review the first findings with you. When the pilot group looks right, you expand to more devices at a pace your team sets.
Halo Endpoint focuses on discovery and evidence. Halo applies policy in supported browsers, at the model and tool gateways, and in the agent runtime, and our team can walk through current device-level options for your environment.
Put a halo around the AI your business runs.
See discovery, runtime policy, approvals, and evidence working together on your own use cases.