About Hilt
Hilt sells a platform that maps data movement across cloud and endpoint infrastructure. Hilt is Data Movement Governance for modern infrastructure. Hilt ties each move to a real person and the job that ran it, then scores it against how the data normally moves. Hilt uses a temporal GNN with proprietary ML models to score every move, and agentic investigation writes the case. Hilt says the most valuable data leaves on credentials that are supposed to have it.[1][2][3][4]
Hilt collectors are agents on customer hosts that produce metadata-level telemetry, shipped as a Linux static binary and a macOS managed agent. Hilt reads metadata only by default: who moved what, from where, to where, and how many bytes. Content-aware inspection in Hilt is a toggle the customer controls. Hilt supports both metadata-only and content inspection data collection. Hilt collectors read process, file, and network activity at the kernel across Linux hosts, VMs, containers, and bare metal. On macOS, Hilt uses a system collector and a managed browser extension to watch the last hop, resolved to the person and process behind it. Hilt device endpoints run kernel collectors on macOS and Windows, plus a managed browser extension across Edge, Safari, Firefox, and Chromium browsers. Hilt presents a move that begins on Linux and ends on macOS as one story, not three logs.[2][3][4][5]
One example rule is to alert when payroll data is uploaded outside the approved payroll job. Hilt can send findings to Slack or email, or export them to any SIEM. Hilt response options are to alert, block the upload, or isolate the host. Hilt observation runs off the data path, and enforcement is opt-in and acts at the endpoint, never inline on a server's high-throughput path.[2]
Hilt can isolate a host at the network from the control plane, quarantined while the rest of the fleet keeps running. Hilt intervention is optional and set per workload; on Linux the response is host quarantine, and on macOS Hilt can block the exfil itself. Whether Hilt alerts or quarantines the host, the customer gets a written case covering what happened, why it is unusual, the evidence, and what was done, in plain language. Hilt response is opt-in and off by default.[4]
Hilt deploys as a single-tenant platform inside the customer's own cloud account. Each customer receives a dedicated deployment in its own AWS account, GCP project, Azure subscription, or Ali Cloud account. Hilt applies a Terraform configuration to create the platform infrastructure in the account the customer nominates. The Hilt dashboard is a web application for investigation, detection rules, and administration, reached over private networking. The Hilt analytics database is a column store that holds normalized telemetry and findings. Hilt does not receive, mirror, or hold out-of-band access to customer events. The single exception is the anonymized data export, a core pipeline that transfers only anonymized statistical features after the customer's review and approval. A Hilt collector has exactly one outbound destination: the ingest endpoint, over HTTPS on TCP/443.[3]
Hilt collectors roll across a fleet, one per host, with no reboot, no inline tap, and nothing in front of the traffic. Hilt collectors run off the data path with no listening ports and one outbound rule, and cannot drop, delay, or alter a packet. Hilt operates the platform components in the customer's account, shipping releases, applying updates, and monitoring platform health. The Hilt auto-update agent runs as a Kubernetes CronJob in the customer's cluster. For airgapped environments, Hilt ships an offline bundle with the full artifact set.[4][5]
Hilt engineers do not hold standing access to the customer's environment. In the Hilt dashboard, teams constrain which hosts a user can see. Hilt records security-relevant actions to the audit log, which administrators manage.[1][2][5][6]
Hilt on the record