TL;DR: Kestrel turns natural-language prompts into deterministic, production-ready workflows that automate incident response, cloud provisioning, CI/CD, security, and developer requests. It supports 30+ integrations and is developer-first, with a CLI, Python SDK, and MCP server so you can manage workflows from where you (and your coding agents) already work. Build your first workflow in <5 mins with $1,000 in usage credits at usekestrel.ai.

https://youtu.be/2y60CgQ-huM

The Problem

Coding agents made every developer faster. But the platform team did not get faster with them.

Developers now ship with Cursor/Claude/Codex and generate infrastructure, deployments, and provisioning requests at a rate no platform team was staffed for.

At illumio I once waited more than three weeks just to get Kafka clusters provisioned for a new service I was building. And provisioning is only part of it. The platform team is also on the hook for incident response when production breaks, repairing the CI/CD glue that ties the deployment pipeline together, platform security, and every "can you just" developer request in Slack.

Most platform teams pick one of two bad options. They hand developers (and their agents) direct access to infra, CI/CD, and cloud accounts, and hope nothing breaks. Or they gatekeep and become the bottleneck every developer has to wait on.

Building a platform that developers actually like has been a luxury only the largest engineering orgs could staff, and an understaffed platform effort usually does more harm than good.

The Solution

See how it works: https://youtu.be/g1dGUsuZgws

Platform teams use Kestrel to automate incident response, cloud provisioning, CI/CD, security, developer requests, and more across their stack. You describe what you want to automate in plain English, and the Workflow Agent builds the whole thing. It draws from 140+ pre-built actions across 30+ integrations, plus custom HTTP APIs and webhooks for anything that isn't built in yet.

Once a workflow is configured it runs deterministically with no LLM deciding what to do at runtime.

You can define workflows that handle self-service developer requests without handing over the keys; scope who can trigger each workflow; fence it to specific clusters, cloud accounts, or repos; and require approval before anything touches production. Developers get a golden path without free rein, and the platform team keeps the guardrails without gatekeeping.

Kestrel analyzes your connected integrations and suggests ready-to-deploy workflows, and the Workflow Assistant turns your existing runbooks and scripts into production-ready automations, so you don't have to start from scratch: https://youtu.be/IOoLb300rqE

Every workflow you build in Kestrel is fully observable, with a dashboard that shows what's running, what's waiting on approval, and what failed; real-time step-level logs; and a failure diagnosis agent that fixes failed workflows for you: https://youtu.be/SEInJ040ZVs

The Kestrel CLI, Python SDK, and MCP server let you and your coding agents build and manage workflows from the terminal and application code.

How It Works

First, connect your platform stack. Kestrel supports 30+ integrations across infrastructure, PaaS, AI compute, observability, on-call, networking, databases, CI/CD, and IaC:

Then, open the Workflow Agent and describe what you want to automate. Here are a few examples:

• “When a PagerDuty alert fires for the production Kubernetes cluster, run root cause analysis, post the findings to #incidents in Slack, create a P1 Jira ticket, and apply the recommended fix after approval in Kestrel.”
• “When a developer submits a cloud provisioning request in Slack, route it to the team lead for approval, provision the AWS resources with Terraform in the infra repo, and notify the developer once the pull request is merged.”

Everything you do in the dashboard is available via the CLI/SDK/MCP. E.g. you can generate a workflow, activate it, and clear an approval from the terminal:

Ask

If you're a platform engineer or eng leader tired of manually responding to developer requests, incidents, or CI/CD failures, you can try it yourself today with a 14-day free trial and $1,000 in usage credits. Use suggested workflows and the Workflow Assistant to build your first platform automations in minutes.

If you want to learn more about what onboarding looks like or dedicated infrastructure/on-prem deployment options, book a demo and I'll walk you through it.

Try Kestrel today