Octopus Deploy’s Dan Garfield on Argo CD, Guardrails, and Octo Doctor

Octopus Deploy’s Dan Garfield on Argo CD, Guardrails, and Octo Doctor | KubeCon NA 2026

Share:

David Marshall | Published: November 12, 2026

Octopus Deploy’s VP of Open Source, Dan Garfield, on Guardrails for Agentic Development, Octo Doctor, and Argo CD Cost Intelligence AI is accelerating development, but the bottleneck is moving downstream.

In this interview filmed at the Octopus Deploy booth at KubeCon + CloudNativeCon North America 2026, VMblog talks with Dan Garfield about how continuous delivery is becoming the next pressure point for teams adopting AI-assisted development.

Garfield explains that organizations first got stuck on Git, then on CI, and are now hitting limits with CD, whether they’ve been relying on Jenkins or another tool. Octopus Deploy, a continuous delivery platform built on Kubernetes and Argo CD, helps teams put guardrails in place, such as production-specific approvals, testing, and validation, that can be templated across projects to scale.

In this video, we cover:

What Octopus Deploy is and how it helps get software onto deployment targets, validate it in staging, and promote it to production
Why AI-driven development produces high output but not always high confidence, and how CD helps close that gap
Octo Doctor, a free, open-source, deterministic (not LLM-based) tool that scans Argo CD instances for bad practices and recommends fixes. It can run from a CLI, a Claude skill, or as a CI step
How the Octopus MCP server helps teams spin up pipelines using their preferred AI tools, and why Garfield says adopters move faster and see more reliable pipelines
Using ephemeral environments to validate rapid, agent-driven changes
Applying global policies, such as requiring a second approval for AI-remediated changes before they reach production
Tight Argo CD integration: triggering syncs, waiting for healthy status, and staging rollouts across apps and environments
Cost intelligence for Argo CD, which surfaces cost data at the application and change level directly in the UI, so teams can spot, for example, a 20% cost increase after a sync
The shift from deploying around 10 or 100 times a day to 500 or even 1,000 times a day

Get started:

Octopus Deploy is free to try, with access to all features, including connecting a Kubernetes cluster, running a deployment, and using the MCP server.