How much does groundcover cost?

Billed per monitored Kubernetes or Linux host on the monthly average. Free is $0 with 12 hour retention, Pro is $30 per host a month billed yearly or $35 billed monthly, with SSO and standard retention, Enterprise is $35 per host a month billed yearly or $40 billed monthly, with RBAC and unlimited retention, and On Premise is $50 per host a month. Agent Mode model calls bill to the customer's own cloud or model provider account, with monthly budgets per organization and per user. Plans are also sold on the AWS, Google Cloud and Azure marketplaces.

What are the best alternatives to groundcover?

Based on documented capability coverage across the Agentic Index 14 point taxonomy, the closest alternatives to groundcover among SRE and DevOps agents are Doctor Droid, Bluebricks, Corelayer, Fabrix.ai. Each is graded on the same public evidence, and no vendor pays for placement.

Does groundcover support MCP and tool calling?

Full under the Agentic Index verification standard. groundcover publishes public evidence of Model Context Protocol (MCP) support, tool calling, and API or SDK extensibility for connecting agents to external tools. Evidence basis: A REST API covers queries over logs, traces, events, metrics and AI spans and the management of dashboards, monitors, silences, notification routes, pipelines, service accounts, RBAC policies and the agent's own instructions, budgets and Sk. Graded from public evidence only against the Agentic Index 14 point capability taxonomy. No vendor pays for placement.

Does groundcover have SOC 2, SSO and role based access control?

Full under the Agentic Index verification standard. groundcover publishes public evidence of enterprise security and identity controls, covering a compliance posture such as SOC 2 or ISO 27001, single sign on (SSO or SAML), and role based access control (RBAC). Evidence basis: Sign in runs through single sign on from the Pro plan, with Okta connected as an OIDC identity provider and its roles mapped in, and Enterprise adds RBAC policies with data scopes that decide which clusters and signals each person or servic. Graded from public evidence only against the Agentic Index 14 point capability taxonomy. No vendor pays for placement.

Can groundcover be self hosted or run in a private VPC?

Full under the Agentic Index verification standard. groundcover publishes public evidence of self hosted or private VPC deployment, along with data residency options for controlling where agent data is processed and stored. Evidence basis: The data plane runs inside the customer's own AWS, Google Cloud or Azure account under the bring your own cloud model, managed by groundcover, so logs, traces and agent conversations stay in the customer's infrastructure. The On Premise edi. Graded from public evidence only against the Agentic Index 14 point capability taxonomy. No vendor pays for placement.

Back to vendors

groundcover

Also known as: groundcover Agent Mode, groundcover AI Mode, groundcover MCP, groundcover.com

Visit site

SRE / DevOps agentindependentVerified 2026-11-09

Entry priceFree at $0 with 12 hour retention. Pro is $30 per host a month billed yearly or $35 billed monthly, Enterprise $35 billed yearly or $40 billed monthly, and On Premise $50, all on the monthly average of monitored hosts.Full pricing detail

groundcover is an eBPF based observability platform that runs inside the customer's own cloud, with Agent Mode as its built in investigator. The agent queries logs, traces, metrics and Kubernetes state, explains what broke, drafts dashboards, monitors and log parsing rules, and acts through GitHub, PagerDuty, Linear, Jira and coding agents such as Cursor. Admins pick the model provider, including Anthropic or OpenAI, and gate each MCP tool. Plans run $30 to $50 per host a month on yearly billing, with a free tier.

groundcover sells observability for Kubernetes and Linux environments built on an eBPF sensor that captures logs, traces, metrics, events and real user data without code changes. Its bring your own cloud architecture keeps the data plane in the customer's AWS, Google Cloud or Azure account, managed by groundcover, and a fully on premises edition runs the UI and backend on the customer's own servers. The company is groundcover, Inc., with an office on HaMasger Street in Tel Aviv, and its customers include Unity, Fiverr, Cargill, Akamai and Cresta.

Agent Mode is the agent layer. Engineers ask it in plain language why a service is failing, and it queries logs, traces, metrics, Kubernetes events and live entities, shows every query and result as an expandable step, and reports a likely cause. It drafts dashboards, monitors, log parsing and drop rules for review, and works from Slack through @groundcover mentions.

Per user connectors let it act on the user's behalf in GitHub, PagerDuty, Linear, Jira and Confluence through Atlassian Rovo, incident.io, Rootly, Pylon and Notion, and it can hand a code fix to Cursor Cloud Agents or Claude Managed Agents. Admins allowlist other remote MCP servers and set each tool to run automatically, wait for approval or stay blocked. Personal and organizational Skills hold team runbooks that activate when a prompt matches them.

Separately, AI Observability captures the customer's own LLM calls and agent workflows through eBPF and OpenTelemetry, with tokens, cost and latency, and an MCP server lets outside agents query groundcover.

Pricing is per monitored host, averaged over the month. Free costs $0 with 12 hours of retention, Pro is $30 per host a month billed yearly or $35 billed monthly, with SSO and up to 30 days of logs and traces, Enterprise is $35 billed yearly or $40 monthly, with RBAC and unlimited retention, and On Premise is $50 with a self hosted UI and isolated authentication. Model calls run on the customer's own Bedrock, Vertex AI, Microsoft Foundry, Anthropic or OpenAI account, with monthly budgets per organization and per user. groundcover holds SOC 2 Type 2 and ISO 27001 and lists HIPAA and PCI DSS on its trust center.

groundcover fits Kubernetes heavy teams that want an investigation agent working on telemetry that never leaves their cloud, with control over the model and over which tools run without approval. It fits less well for teams that want the agent to start investigations on its own when an alert fires, need a library of prebuilt agents, or run mostly outside Kubernetes and Linux hosts.

Vendor details

Canonical URL

https://www.groundcover.com

Subcategory

Investigation agent inside a bring your own cloud observability platform, with observability for the customer's own LLM apps and agents

Funding status

Privately held. groundcover, Inc., with an office in Tel Aviv.

Company status

independent

Use cases & customers

Primary use cases

investigating errors and latency across logs, traces, metrics and Kubernetes events in plain languagedrafting dashboards, monitors and log parsing or drop rules from a descriptionturning an investigation into a Linear or Jira ticket or a GitHub changehanding a code fix to Cursor Cloud Agents or Claude Managed Agentstracing the customer's own LLM calls and agent workflows with token and cost data

Target customers

Kubernetes platform and SRE teams that want observability data to stay in their own cloud accountengineering teams that want an agent to investigate incidents and hand fixes to coding agentsregulated companies that need an on premises observability stack with an agent on itAI product teams that want their LLM calls and agent workflows traced without code changes

Deployment options

Bring your own cloud on AWS, Google Cloud or Azure, managed by groundcoverFully on premises with self hosted backend and UIgroundcover Cloud

Integrations

Agent connectors for GitHub, PagerDuty, Linear, Atlassian Rovo (Jira and Confluence), incident.io, Rootly, Pylon, Notion, Slack, Cursor Cloud Agents and Claude Managed Agents, plus allowlisted remote MCP servers; notification destinations for Slack, PagerDuty, OpsGenie, Microsoft Teams, incident.io, Jira and webhooks; model providers Amazon Bedrock, Google Cloud Vertex AI, Microsoft Foundry, Anthropic, OpenAI and compatible gateways such as LiteLLM and Bifrost; OpenTelemetry, Prometheus and Datadog agent ingest; Go, Python and TypeScript SDKs, Terraform provider and an MCP server.

In practice

A checkout service starts returning 500 errors. An engineer mentions @groundcover in the incident channel, the agent pulls the failing traces and recent deploy events, finds a feature flag left on in the payments service, and posts the evidence in the thread before opening a Linear ticket with the root cause.

A platform team allowlists its internal deployment MCP server and sets the rollback tool to need approval. When the agent proposes a rollback during an investigation, the on call engineer approves it in the conversation, while read only tools keep running on their own.

An AI team running a support agent on OpenAI and Anthropic sees every model call appear in AI Observability with tokens, cost and latency, then adds OpenTelemetry GenAI attributes to group spans by conversation and agent name.

Sources & related URLs

Related / legacy domains

https://docs.groundcover.com/automate/agent-mode https://www.groundcover.com/pricing https://trust.groundcover.com/ https://docs.groundcover.com/automate/connectors https://www.groundcover.com/product/agent-mode https://www.groundcover.com/ai-observability

Research sources

Agentic Index coverage score

11.0 / 14 capabilities · 79%

Integrations & Tool Calling	Full
Agent Mode acts in outside tools through connectors each user authorizes with OAuth or a personal token: it opens and updates GitHub issues and pull requests, manages PagerDuty incidents and schedules, creates Linear issues, reads and updates Jira and Confluence through Atlassian Rovo, and works incidents in incident.io and Rootly, support tickets in Pylon and pages in Notion. Admins enable each connector for the organization, and they can allowlist any other remote MCP server with token or OAuth sign in, which adds that server's tools to the agent. Sourcedocs.groundcover.com/automate/connectors.mdread 2026-11-09
Workflow Orchestration	Full
An investigation runs as a chain of queries, lookups and actions that the agent plans itself, and users can queue further prompts or steer a run in progress without restarting it. When a fix needs code, the agent hands the task to a second agent, Cursor Cloud Agents or a Claude Managed Agent with its own environment, using the user's connected credentials. Skills package multistep procedures such as a payments triage that anchors the timeline, checks blast radius, correlates deploys and traces a failing request in order. Alert workflows route notifications rather than running agents. Sourcedocs.groundcover.com/automate/agent-mode/getting-started.mdread 2026-11-09
Knowledge Grounding & RAG	Full
Teams keep their runbooks, investigation playbooks and naming conventions as Skills, personal or organization wide, each with a When to use description the agent reads to pull the relevant ones into a conversation automatically. Skills are versioned, admins manage the shared ones, and the agent can draft a new Skill from a finished investigation for review. Through connectors it also searches Notion pages and Confluence spaces, and every answer draws on the customer's own logs, traces, metrics, Kubernetes events and live entities queried at run time. Sourcedocs.groundcover.com/automate/agent-mode/skills.mdread 2026-11-09
Human Oversight & Guardrails	Full
Every tool on an allowlisted MCP server is set to run automatically, wait for the user's approval on each call, or stay blocked, with Needs Approval as the default. The admin's setting is a ceiling, so a user can tighten a tool but never loosen it, and new tools inherit the admin default. Inside groundcover the agent produces dashboards and monitors as drafts a person saves. Its queries and actions run with the user's own RBAC permissions, admins decide which connectors exist at all, and monthly budgets cap spend per organization and per user. Sourcedocs.groundcover.com/automate/connectors/mcp.mdread 2026-11-09
Security, Identity & Governance	Full
Sign in runs through single sign on from the Pro plan, with Okta connected as an OIDC identity provider and its roles mapped in, and Enterprise adds RBAC policies with data scopes that decide which clusters and signals each person or service account can see. Service accounts hold their own API keys and inherit their policies, and only admins can change AI settings. The trust center lists SOC 2 Type 2, ISO/IEC 27001:2022, HIPAA, PCI DSS, GDPR and CCPA, with the SOC 2 report and a penetration test report on request. Sourcetrust.groundcover.comread 2026-11-09
Observability & Auditability	Full
Each step the agent takes, every query, lookup or action, appears as an expandable item showing the query it ran, its parameters and the raw results, and the agent names the Skills it used on each message. Conversations stay for 30 days of inactivity and can be shared as read only links. For the customer's own AI systems, AI Observability captures LLM calls through eBPF with model, tokens, cost, latency and full prompts and responses, and OpenTelemetry GenAI spans add agent names, tool chains and conversation threads. Sourcedocs.groundcover.com/automate/agent-mode/getting-started.mdread 2026-11-09
Memory & State Persistence	Partial
Each conversation keeps its own message history, which gives the agent context for follow ups and multistep investigations; sessions expire after 30 days without activity and their history is deleted, and in bring your own cloud deployments it sits in a database inside the customer's own infrastructure. History stays with the user who started it, and a teammate can only continue it by forking a shared link. Personal Instructions apply to every chat a user starts, but the agent carries nothing it learned from one conversation into the next. Sourcedocs.groundcover.com/automate/agent-mode/privacy-and-security.mdread 2026-11-09
Deployment & Data Residency	Full
The data plane runs inside the customer's own AWS, Google Cloud or Azure account under the bring your own cloud model, managed by groundcover, so logs, traces and agent conversations stay in the customer's infrastructure. The On Premise edition puts the backend and UI on the customer's own cloud or physical servers with isolated authentication. Agent model calls go to Amazon Bedrock, Vertex AI or Microsoft Foundry in the customer's cloud account, though Bedrock cross Region inference can route outside the cluster's region. Sourcedocs.groundcover.com/automate/agent-mode/privacy-and-security.mdread 2026-11-09
Prebuilt Agents / Templates / Packs	Partial
New conversations open with template cards for error investigation, performance analysis and infrastructure health, and a page of example prompts covers common scenarios. Beyond that, teams build their own Skills, writing them by hand or asking the agent to turn a finished investigation into one, and admins share them across the organization. The monitor and dashboard catalogs offer prebuilt monitors and dashboards rather than agents, and Agent Mode is a single assistant rather than a set of separate agents. Sourcedocs.groundcover.com/automate/agent-mode/getting-started.mdread 2026-11-09
Triggers & Channel Coverage	Partial
Work reaches the agent when a person asks: from the chat panel in the groundcover app, from an @groundcover mention in Slack, from embedded prompts in the dashboard, widget and monitor builders, or from an Investigate with Agent link on a Slack alert, which opens the issue with the agent ready. Outside agents can also call it through MCP tools that start an asynchronous groundcover agent run. Monitors fire notification routes to Slack, PagerDuty, OpsGenie, Teams and webhooks, but a firing issue does not start an investigation on its own, and there is no schedule for agent runs. Sourcedocs.groundcover.com/automate/agent-mode/getting-started.mdread 2026-11-09
Model Flexibility & Routing	Full
Admins choose the provider behind the agent for each backend: the deployment's own Amazon Bedrock, Vertex AI or Microsoft Foundry, Anthropic or OpenAI on the customer's key, or an Anthropic or OpenAI compatible gateway such as LiteLLM or Bifrost. On Bedrock they enable models from a list or by model ID and set a default, while other workspaces default to Claude Sonnet 4.6. Each user then picks among the enabled models, and a reasoning effort level where offered, and can switch models between messages in the same conversation. Sourcedocs.groundcover.com/automate/agent-mode/configuring-settings.mdread 2026-11-09
APIs / SDKs / MCP Extensibility	Full
A REST API covers queries over logs, traces, events, metrics and AI spans and the management of dashboards, monitors, silences, notification routes, pipelines, service accounts, RBAC policies and the agent's own instructions, budgets and Skills. Official Go, Python and TypeScript SDKs wrap it with service account API keys, a Terraform provider manages resources as code, and the groundcover MCP server lets Cursor, Claude Desktop and other agents query data or start a groundcover agent run. Sourcedocs.groundcover.com/reference/remote-access-and-apis/sdks.mdread 2026-11-09
Testing, Debugging & Optimization	Partial
AI Observability helps teams debug their own LLM applications and agents: every model call carries model, token counts, cost and latency, prompts and responses sit in the trace drawer, and spans group by conversation and agent name for comparison over time. It records rather than scores, with no test sets, judges or quality metrics. For Agent Mode itself, users verify the agent's work by expanding each step, and refine a Skill and try again when the agent misses. Sourcedocs.groundcover.com/investigate/ai-observability/ai-observability.mdread 2026-11-09
Browser / Computer-use	Not documented
The agent never drives a browser or desktop. It reaches systems through telemetry queries, connectors and MCP tools, and its navigation feature only opens the right page of the groundcover app for the user with filters applied. Synthetic checks and real user monitoring watch the customer's sites and apps as monitoring products, and screenshots a user attaches are read as context rather than operated. Sourcedocs.groundcover.com/automate/agent-mode.mdread 2026-11-09

The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded

Pricing

Free at $0 with 12 hour retention. Pro is $30 per host a month billed yearly or $35 billed monthly, Enterprise $35 billed yearly or $40 billed monthly, and On Premise $50, all on the monthly average of monitored hosts.

Per monitored Kubernetes or Linux host, averaged over the month, with agent model usage billed by the customer's own cloud or model provider

Free tierTrial available

Included quota

Plans do not meter data volume: Pro keeps up to 30 days of logs, traces and events and 13 months of metrics, Enterprise has unlimited retention, and Free keeps 12 hours.

What is public

Per host prices for Free, Pro, Enterprise and On Premise under both monthly and yearly billing, retention limits and what each plan adds, and the deployment mode each plan runs in.

Billing mechanics

groundcover bills on the monthly average of monitored Kubernetes nodes and Linux hosts, while agent model calls bill through the customer's own provider account with budgets set in groundcover.

Cost watchouts

Agent Mode model calls bill to the customer's own cloud or model provider account on top of the per host fee, and BYOC runs the data plane on the customer's own cloud infrastructure, which carries its own compute and storage cost.

Variable cost rationale

The platform fee follows the average number of monitored hosts, not data volume, but agent use adds model charges on the customer's own provider account, which grow with how often people investigate and which models they pick; admins can cap that spend with monthly budgets.

Additional watchouts

RBAC needs Enterprise, SSO needs Pro, and the agent's model usage is a separate cost on the customer's own provider account.

Overage / add-ons

No data volume overage; the host count is averaged over the month so short peaks do not raise the bill.

Sales call required

Mixed (some tiers require a call)

Free / trial

Free plan at $0 with 12 hour retention and community support, no card required; Pro and Enterprise start with a free trial.

Lowest paid plan

Pro at $35 per host a month, or $30 billed yearly

Commercial notes

groundcover, Inc. sells the platform direct and through the AWS, Google Cloud and Azure marketplaces, with custom billing plans through sales.

Key ambiguities

The pricing page lists no separate charge for Agent Mode, and On Premise carries a single $50 figure with no monthly or yearly choice.

Cancellation / refund

Free needs no card; Pro and Enterprise can be billed monthly or yearly, with yearly billing $5 a host cheaper each month, and custom terms come through sales or the cloud marketplaces.

Support SLA / resale

Free has Slack community support, Pro standard support, Enterprise premium support, and On Premise dedicated support from the groundcover team; plans are also sold through the AWS, Google Cloud and Azure marketplaces.

Missing data

The cost of running the data plane in the customer's own cloud.

Agentic Index verified 2026-11-09

Official pricing page Docs

Related vendors

Alternatives to groundcover

The closest documented capability profiles to groundcover among SRE and DevOps agents tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.

Doctor Droid11.5 / 14Fuller documented coverage on Triggers & Channel Coverage
Bluebricks10.5 / 14Fuller documented coverage on Triggers & Channel Coverage
Corelayer11.5 / 14Fuller documented coverage on Memory & State Persistence and Triggers & Channel Coverage
Fabrix.ai12.5 / 14Fuller documented coverage on Prebuilt Agents, Templates & Packs and Triggers & Channel Coverage
Honeycomb10.5 / 14Fuller documented coverage on Triggers & Channel Coverage
Kubiya11.5 / 14Fuller documented coverage on Prebuilt Agents, Templates & Packs and Triggers & Channel Coverage

Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded

Compare groundcover with other vendors →Compare all SRE and DevOps agents →See where groundcover ranks →

Data confidence: high