Skip to main content

​
What You’ll Learn

Assessment analyzes your cloud resources against the AWS Well-Architected Framework. You can select a single resource or multiple, one pillar or all six. Start small to learn the workflow, then scale up to surface patterns across your infrastructure.
Assessment requires the Advanced plan. Only resources that appear in your Resources inventory can be assessed — make sure discovery has run first.
1

Navigate to Assessment

Go to Infrastructure > Assessment in your workspace.
2

Select Resources

Choose which resources to assess. For your first run, start with one resource — a good candidate is:
  • An EC2 instance running a production workload
  • An RDS database you suspect is over-provisioned
  • An S3 bucket with unknown access patterns
As you get comfortable, scale up by selecting multiple resources using filters:
FilterExamples
Service typeAll EC2 instances, all RDS databases
TagsEnvironment: production, Team: backend
Regionus-east-1, all US regions
You can select resources individually or click Select All on a filtered list.
3

Choose Pillars

Select which Well-Architected pillars to evaluate:
PillarWhat It Checks
Cost OptimizationRight-sizing, reserved capacity, idle spend
SecurityEncryption, access control, compliance
ReliabilityBackup, redundancy, fault tolerance
Performance EfficiencyThroughput, latency, resource utilization
Operational ExcellenceMonitoring, automation, runbook coverage
SustainabilityResource efficiency, carbon footprint
Start with one pillar — Cost Optimization or Security typically surface the most actionable findings. Add more pillars as you scale up.
More resources and pillars means longer assessment time. A single resource with one pillar takes 1-2 minutes. Scale gradually.
4

Run the Assessment

Click Run Assessment. The AI agent:
  1. Collects current resource configuration
  2. Analyzes metrics and usage patterns
  3. Evaluates against best practices for each pillar
  4. Generates findings with severity ratings
You’ll see a progress indicator as resources are analyzed.
5

Review Results

The analytics dashboard shows:
  • Findings by pillar: How many issues were found in each category
  • Severity breakdown: Critical, High, Medium, Low distribution
  • Potential savings: Estimated cost reduction if recommendations are implemented
When assessing multiple resources, the dashboard also shows:
  • Top affected resources: Which resources need the most attention
  • Pattern detection: Systemic issues across resources (e.g., 80% of EC2 instances over-provisioned, no S3 buckets with versioning enabled)
6

Explore Recommendations

Each finding includes a recommendation with:
  • Description: What needs to change and why
  • Impact level: High, Medium, Low
  • Effort: How much work to implement
  • Risk: What could go wrong during implementation
For each recommendation, you have four actions:
  1. Impact Analytics: See projected impact before making changes
  2. Generate Guidelines: Create implementation documentation
  3. Custom Prompt: Ask agents follow-up questions about the finding
  4. Implement: Apply the fix with agent assistance
When reviewing multiple resources, sort by Potential savings, Severity, or Effort to prioritize. Focus on systemic issues — fixing one policy can resolve findings across many resources.
7

Save to Plan

Recommendations are saved as drafts by default. To track them:
  1. Select the recommendations you want to act on
  2. Click Save to Plan (batch select works for multiple recommendations)
  3. Assign priority and timeline on the Plan page
  4. Track implementation status over time
The Plan page gives you a central view of all pending and completed optimizations across your infrastructure.

​
Example: First Assessment

Resource: i-0abc123def456 (m5.2xlarge, us-east-1) Pillar: Cost Optimization Findings:
  • Instance is over-provisioned — average CPU at 12% over 30 days
  • No Reserved Instance or Savings Plan coverage
  • Recommendation: Downsize to m5.large, saving ~$180/month
Once you’ve seen how it works, run the same pillar across all your EC2 instances to see if over-provisioning is a systemic pattern.

​
Tips

  • Start small: 1 resource + 1 pillar for your first run, then scale up
  • Use filters to group: Tag-based filtering (environment, team) makes multi-resource assessments much more targeted
  • Check Impact Analytics first: Always review projected impact before clicking Implement
  • Run periodically: Weekly assessments catch drift and new issues over time
  • Use Anna for summaries: After a large assessment, ask @anna #report summarize the top 5 issues and create an action plan

​
Next Step

Incident Response

Set up automated incident detection and root cause analysis