Skip to main content

Module 2 Quiz: Platform AI

Five questions covering AWS AI services, their capabilities, and the gap custom agents fill.

Question 1

Which AWS AI service is free to use in the web console for cost analysis — with no per-request charges?

Answer

Cost Explorer — the web console is free. The Cost Explorer API charges $0.01 per request, but browsing and analyzing costs through the AWS Console interface incurs no charges.

Question 2

What is the main limitation of CloudWatch Anomaly Detection compared to a custom agent analyzing the same alarm?

Answer

CloudWatch Anomaly Detection detects that a metric deviated from baseline. It fires an alert. It cannot:

  • Cross-reference the anomaly with your deployment pipeline
  • Check whether this pattern matches a known failure mode in your runbook
  • Execute investigation steps (query related metrics, check recent changes)
  • Produce a structured diagnosis

A custom agent with domain context (runbook, topology, deployment history) can do all of these — detection is just the first step.

Question 3

Amazon Q Developer's free tier requires what type of login — and why is the distinction important?

Answer

Q Developer's free tier uses an AWS Builder ID — a personal identity that is free to create and requires no AWS account, no credit card, and no billing setup.

The distinction matters because participants who don't have an AWS account (or don't want to use theirs) can still complete the Q Developer lab. AWS Builder ID is separate from IAM credentials and incurs no charges.

Question 4

What is vendor lock-in in the context of platform AI features, and why does it matter for DevOps teams?

Answer

Vendor lock-in means your operational AI intelligence is embedded in one cloud provider's tooling. If you use CloudWatch Anomaly Detection, DevOps Guru, and Cost Explorer, that intelligence only covers your AWS workloads.

It matters because:

  • Multi-cloud and hybrid teams can't get a unified view
  • If you migrate providers, you lose your anomaly detection history and configuration
  • You have no leverage to bring your operational knowledge to a different platform

Custom agents built on a provider-agnostic framework (Hermes) encode operational knowledge in portable SKILL.md files that work regardless of which cloud or tools you use.

Question 5

A team has CloudWatch Anomaly Detection alerting on their API Gateway latency. The alert fires. What will platform AI handle, and what will still be manual (or require a custom agent)?

Answer

Platform AI handles: Detecting the anomaly and sending the alert.

Still manual (or requires a custom agent):

  • Querying related metrics (Lambda duration, downstream service latency, DynamoDB read capacity)
  • Checking the deployment pipeline for recent changes
  • Correlating the pattern with previous incidents
  • Following the runbook to narrow down root cause
  • Deciding whether to roll back, scale out, or escalate
  • Creating a structured incident ticket with the diagnosis

The alert tells you something is wrong. Everything after that is investigation — and investigation requires context that platform AI doesn't have.

Question 6

True or False: AWS DevOps Guru can execute remediation actions when it detects an operational issue.

Answer

False. DevOps Guru detects and surfaces operational anomalies across your AWS services. It creates OpsItems in Systems Manager OpsCenter, but it does NOT take remediation action.

Remediation still requires either a human engineer or a custom agent with the appropriate tools and governance controls. Detection ≠ action.

Last updated on