Productized and supported, not DIY
Hyground ships as a supported enterprise platform with a managed release cadence. HolmesGPT is an open-source project you assemble, host, and operate yourself, or you adopt Robusta's SaaS for the extras.
Hyground vs Robusta / HolmesGPT
A governed operations platform, not an open-source agent you assemble and run yourself
HolmesGPT is an open-source investigation agent, read-only and LLM-agnostic, with an optional Robusta SaaS for history and change tracking. Hyground shares the in-cluster, read-only, bring-your-own-LLM architecture but ships it as a governed, supported platform: a central gateway with platform RBAC and audit, plus workflow automation across your operations.
Same architecture, different product
HolmesGPT, a CNCF Sandbox project built by Robusta and maintained with Microsoft, runs an agentic loop over your observability toolsets, read-only and RBAC-aware, deployable as a CLI, an HTTP API, or an in-cluster operator; the richer history, change tracking, and dashboards live in Robusta's SaaS. Hyground takes the same in-cluster, read-only, bring-your-own-LLM stance and adds enterprise governance, orchestration across observability and ITSM, and reusable, schedulable skills.
Architecture
Where Hyground differs
Six choices that separate an open-source building block from a governed, supported platform.
Central gateway with platform RBAC and audit
Hyground routes every tool call through one in-cluster gateway with platform-level RBAC and a full audit trail. HolmesGPT inherits Kubernetes RBAC per service account; cross-tool governance and a unified audit are left to you or to Robusta's platform.
Operations beyond troubleshooting
Hyground turns investigations into reusable skills you can schedule and automate across operations. HolmesGPT focuses on investigation and root cause, with a proactive Operator mode and fix pull requests.
Bidirectional ITSM and OSS connectors, first-party
Hyground ships supported first-party connectors, including bidirectional Jira and ServiceNow. HolmesGPT integrates many toolsets that you configure and maintain yourself.
Shared strengths, kept
Both are read-only by default and LLM-agnostic across Azure OpenAI, Anthropic, AWS Bedrock, Google Gemini, and self-hosted models. If you value those, Hyground keeps them inside a governed platform.
Procurement- and compliance-ready
Hyground is built for enterprise procurement and compliance, with sovereignty, audit, and support agreements. An open-source agent plus a SaaS add-on is a different acquisition and accountability model.
Decision
When each tool fits
HolmesGPT and Hyground share an architecture but sit at different points on the build-versus-buy line: an open-source agent, and a governed platform.
Choose Hyground when
You want the in-cluster, read-only, bring-your-own-LLM approach delivered as a supported, governed platform with audit, ITSM, workflow automation, and procurement-ready compliance.
Choose Robusta / HolmesGPT when
You want a free, open-source troubleshooting agent, have the team to run and integrate it, and are happy to add Robusta's SaaS for history and change tracking.
See Hyground in action
Try the sandbox or schedule a demo and watch the agent investigate against a real cluster.