Kestrel

What it does

Kestrel is an AI-powered platform that helps platform engineering teams automate complex operational workflows. Using natural language, users describe the automation they need—such as incident response, cloud provisioning, CI/CD pipeline orchestration, or developer requests—and Kestrel's AI agent generates a deterministic workflow graph. These workflows are composed of 140+ pre-built actions across 30+ integrations (including Kubernetes, AWS, Vercel, PagerDuty, Slack, Jira, GitHub, and more) and can be further refined with a visual DAG editor, CLI, Python SDK, or MCP server. Once deployed, workflows execute reliably every time, with no hallucinations or surprises, because the AI builds the workflow but the execution runs the exact code defined.

Key capabilities include:

• Workflow Agent: Convert plain English descriptions into production-ready workflow graphs.
• Deterministic Execution: Workflows run the same code every time, ensuring predictable outcomes.
• Access Control: RBAC per workflow, approval gates (in-platform, Slack-based, or PR-based), and a full audit trail.
• Observability: Real-time dashboard with step-level logs, execution history, latency metrics, and failure alerts.
• Developer Tools: Manage workflows from CLI, Python SDK, or MCP server for integration with AI coding assistants.
• Suggested Workflows: AI analyzes connected integrations and suggests ready-to-use workflows.

Who it is for

Kestrel is built for platform engineering teams, DevOps engineers, SREs, and infrastructure teams who need to automate repetitive, multi-step operational tasks across their stack. It is also useful for developers who want self-serve infrastructure provisioning with guardrails, and for security teams that need automated incident response and threat containment. The product is designed for organizations that already use tools like Kubernetes, AWS, PagerDuty, Slack, Jira, and GitHub, and want to connect them into automated, auditable workflows.

Why it matters

Platform engineering teams often spend significant time building and maintaining custom automation scripts, glue code, and runbooks. These manual efforts are error-prone, hard to scale, and difficult to audit. Kestrel reduces this burden by letting teams describe automations in natural language and instantly get a deterministic, production-ready workflow. The deterministic execution model eliminates the risk of AI hallucinations in production, while the built-in access controls and audit trail ensure compliance and security. By automating incident response, cloud provisioning, CI/CD pipelines, and developer requests, Kestrel helps teams reduce mean time to resolution (MTTR), improve developer productivity, and enforce consistent operational practices.

Launch signal

Kestrel launched on Y Combinator and is backed by Y Combinator. The website prominently features this backing and a "We launched on Y Combinator" callout. The product is available as a cloud service or self-hosted, with a 14-day free trial (no credit card required). Pricing is usage-based for workflows (pay per execution, with blocks priced at $0.10 for light, $0.20 for regular, $0.40 for heavy) and per-cluster/per-cloud-account for incident response ($200/month for starter tier). The platform claims SOC 2 compliance, tenant isolation, and read-only-by-default permissions.

Brand and naming

The name "Kestrel" evokes speed, precision, and a bird of prey's keen vision—fitting for a tool that rapidly detects and responds to incidents. The tagline "AI Agents for Platform Engineering" clearly positions the product in the DevOps/AIOps space. The brand emphasizes determinism and reliability, countering common fears about AI unpredictability in production environments.