[WIP] DOCS-14116

content/en/bits_ai/infrastructure.md

@@ -0,0 +1,5 @@
+---
+title: Bits Infrastructure
+---
+
+{{< include-markdown "infrastructure/bits_infrastructure" >}}

content/en/infrastructure/bits_infrastructure.md

@@ -0,0 +1,125 @@
+---
+title: Bits Infrastructure Operations
+description: Use Bits Infrastructure Operations to detect, prioritize, and automatically remediate infrastructure issues.
+---
+
+{{< callout url="https://www.datadoghq.com/product-preview/bits-infrastructure-operations/"
+ btn_hidden="false" header="Join the Preview!">}}
+Bits Infrastructure Operations is in Preview. Use this form to submit your request today.
+{{< /callout >}}
+
+Bits Infrastructure Operations detects infrastructure issues, ranks them by business impact, and executes or proposes remediations. Actions are gated by a configurable guardrail policy and logged to an audit trail. Bits Infrastructure Operations is accessible from the Datadog app, Slack, and the terminal or IDE via the MCP server.
+
+To get started, explore [{{< ui >}}Open Issues{{< /ui >}}][1] in Datadog. See the [{{< ui >}}History{{< /ui >}}][3] tab for your audit trail.
+
+{{< img src="bits_ai/infra/issue_list.png" alt="Your image description" style="width:100%;" >}}
+
+## Using Bits Infrastructure across Datadog
+
+### Kubernetes
+
+Bits Infrastructure Operations detects the following Kubernetes issue types:
+
+- OOMKilled pods
+- CrashLoopBackOff
+- ImagePullBackOff
+- HPA at maxReplicas
+- CPU/memory limits misconfigured
+
+Each detected issue includes a confidence score, a summary of affected services and blast radius, and one or more proposed remediation actions. Depending on your guardrail policy, actions are either executed automatically or routed to an approval workflow.
+After a remediation executes, Bits Infrastructure Operations displays before-and-after signals (replica count, p99 latency, error rate) and surfaces a {{< ui >}}Fix in IaC{{< /ui >}} option to open a Terraform PR for the change.
+
+### Hosts
+
+Bits Infrastructure Operations detects host issues including CPU saturation, memory pressure, and failed EC2 status checks.
+
+### Serverless
+
+Bits Infrastructure Operations detects serverless issues including high Lambda error rates, slow cold starts, and permission misconfigurations.
+
+### ECS
+
+Bits Infrastructure Operations detects ECS task-level issues and proposes remediation actions.
+
+## Integrations
+
+### Slack
+
+Bits Infrastructure Operations sends two types of Slack messages:
+
+- **Digest**: A summary of issues that were self-healed automatically, sent on a configurable schedule. Each entry links to the corresponding audit trail entry.
+- **Approval card**: When an action requires human approval, an approval card is posted to Slack. Approvers can approve or reject directly from Slack without opening the Datadog app.
+
+### MCP server
+
+Bits Infrastructure Operations exposes MCP tools for use with Claude Code or any MCP-compatible client. You can list open issues, retrieve issue details, and approve or reject remediations from the terminal.
+
+## Guardrails
+
+Guardrails define which actions Bits Infrastructure Operations may execute automatically and which require human approval. Policies are evaluated per action based on confidence score, action type, and resource scope.
+All actions—whether auto-approved or manually approved—are recorded in the audit trail.
+
+### User approval
+
+Actions that do not meet your auto-approve policy are routed to an approval workflow. The approval card displays:
+
+- The detected issue
+- The proposed action and confidence score
+- The reasoning behind the confidence score
+- {{< ui >}}Approve{{< /ui >}} and {{< ui >}}Reject{{< /ui >}} controls
+
+Approved actions execute immediately. A rollback option is available for a configurable window after execution.
+
+### Bits auto-approved
+
+Actions that meet all auto-approve criteria in your guardrail policy execute without human review. A notification is sent (in-app or through Slack) for each autonomous action taken.
+
+After the same action is manually approved a configurable number of times, Bits Infrastructure Operations prompts you to add it to your auto-approve policy.
+
+## Remediation actions
+
+| Action | Description | 
+| ------ | ----------- |
+| Restart | Rolling restart of a deployment, pod, or process |
+| Rollback | Revert a deployment to the previous version |
+| Scale | Adjust replica count, HPA maxReplicas, or resource limits/requests |
+| Config change | Update a manifest, environment variable, or cloud resource configuration |
+| IaC PR | Open a Terraform or Helm PR to persist a configuration change |
+| Create monitor | Add a Datadog monitor for the detected failure mode |
+
+When an action cannot be executed automatically, Bits Infrastructure Operations provides a recommended next step.
+
+## Get started
+
+{{< callout url=#
+ btn_hidden="false" header="Join the Preview!">}}
+Bits Infrastructure Operations is in Preview. Use this form to submit your request today.
+{{< /callout >}}
+
+1. In Datadog, go to [{{< ui >}}Open Issues{{< /ui >}}][1].
+2. Configure [{{< ui >}}Guardrails{{< /ui >}}][2].
+3. Connect Slack to receive approval cards and digests.
+4. Optionally, install the Datadog MCP server to use Bits Infrastructure Operations from Claude Code or your IDE.
+
+## Example workflows
+
+### Incident response
+
+1. A monitor fires on checkout-api (5xx errors, elevated p99 latency).
+1. Bits SRE determines the checkout HPA has hit maxReplicas.
+1. Bits Infrastructure Operations surfaces an approval card in the {{< ui >}}Open Issues{{< /ui >}} list and in Slack, showing the proposed HPA scale-up and its confidence score. The action requires approval because it targets a production namespace.
+1. An on-call engineer approves from Slack.
+1. Bits Infrastructure Operations executes the change and displays before-and-after metrics (replica count, p99 latency, error rate).
+1. Bits Infrastructure Operations opens a Terraform PR to raise the maxReplicas cap and creates a monitor for the failure mode.
+1. The detection, approval, action, and post-fix state are recorded in the audit trail.
+
+### Proactive remediation
+
+1. An engineer opens the {{< ui >}}Open Issues{{< /ui >}} page to review open issues.
+1. Bits Infrastructure Operations displays a ranked list that includes a TLS certificate expiring on a service endpoint in six days.
+1. The engineer selects the issue. Bits Infrastructure Operations shows the expiry date, affected service, and a proposed remediation (certificate rotation or an IaC PR to automate future rotation).
+1. The engineer approves. Bits Infrastructure Operations executes the rotation, validates the new certificate is in place, and records the action.
+
+[1]: https://app.datadoghq.com/infrastructure/health/open-issues
+[2]: https://app.datadoghq.com/infrastructure/health/guardrails
+[3]: https://app.datadoghq.com/infrastructure/health/history