Azure Automation

What stands out the most about Azure Automation is how effectively it combines enterprise level security, operational reliability & deep native integration within the Azure Cloud ecosystem to support large scale and repeatable automations. From a security standpoint, the native support for managed identities in our global tenant is a huge impacting factor. We're able to run the automation without embedding the credentials, reducing the risk of mismanagement of identity including theft and the operational overhead. I personally write PowerShell scripts for my daily tasks and use Azure Automation to perform IAM related operations, however, our infrastructure engineers as well as our staff engineers are able to perform infrastructure life cycle operations with fully autonomous python and terraform runbooks which highlights Azure Automation's runbook driven orchestration platform. Once established, these runbooks are highly reliable for us for Business as usual operations, reducing the burden of work from our engineers without the need of an AI agent. It definitely has the most reliable and flexible integration around native Azure services as well as around scheduling and execution engine. Automation may feel like a transitional or legacy leaning service within the ecosystem, however, Microsoft investing in newer automation continuously has been the backbone in our decision making.

Extremely powerful and native integration with other Azure based software platforms.

It's great for automating maintenance, deployments, and patching without having to rely on third-party scripts. The runbooks are flexible - we can use PowerShell or Python, which makes it easier to adapt existing scripts - and the hybrid workers let us handle on-prem and cloud tasks from one place. Once scheduled jobs are in place, they're very reliable, and it's nice not having to babysit routine tasks anymore. The integration with Azure Monitor and alerts also gives us decent visibility into what's happening behind the scenes.

Although this is best suited for structured, scheduled automations, it can feel less natural for highly event driven or real time orchestration scenarios. As my organization is adopting more reactive agentic and automation patterns, clearer guidance or tighter integration with complementary Azure services would help teams design end-to-end workflows without needing to piece together multiple tools. In larger environments, the runbooks need some governance as autonomous footprints are growing in my organization, and our team is investing in our own standards for versioning and ownership. While this is manageable, stronger native capabilities around runbook organization and visibility could further simplify long term maintenance. While runbook authoring in Python & PowerShell is strong, developing & debugging runbooks feels more operational than developer friendly. Enhancements around inline debugging, richer error visibility and smoother local to cloud development workflows would make it easier for my team to iterate quickly and reduce effort & time on running diagnostics.

It's overly complex and relies heavily on knowledge of PowerShell and Python. Also, as far as cold runbooks are concerned, we found it to be rather slow and hence, not applicable to real-time automation. Other drawbacks include lack of native migration tools towards other platforms.

The main frustration is the setup experience. It takes time to configure permissions and link resources correctly, and error messages aren't always very clear when something fails. The interface feels a bit dated and clunky when managing a large number of runbooks, and the logging system could be more detailed for troubleshooting. Pricing can also get tricky to estimate when scaling up hybrid workers, and job queue times sometimes vary more than we'd like.