CloudConsole

CloudConsole · AI Explain

Why did this alarm fire? Ask AI, in your own AWS account

Tap Explain on a CloudWatch alarm and CloudConsole gathers the evidence, from the metric to the deploy before it, and asks Claude on Amazon Bedrock in your own AWS account what happened. Nothing goes through us, and it costs about a cent, billed by AWS to your account.

Explain sheet on checkout-api-5xx: what happened, the likely cause with high confidence, five pieces of evidence and Roll back to revision 41

From alarm to likely cause

On any alarm that isn’t OK, tap Explain this alarm. The answer says what happened, the likely cause with a confidence level, the evidence behind it and the next steps. Every piece of evidence is numbered and opens the screen it came from: the deploy, the target group, the metric. The model may only cite evidence the app actually collected.

The evidence it reads

The alarm, its history and its metric over the last 90 minutes; What changed? (CloudTrail writes, pipeline runs, ECS deployments, GitHub runs); and, depending on the alarm, the Lambda function, the Auto Scaling group, the load balancer’s target health or ElastiCache events. Each source has a few seconds and can fail on its own without stopping the answer.

In your AWS account, not ours

The request is signed on your phone with the account’s own credentials and goes straight to Amazon Bedrock in the region you pick. There is no CloudConcur server and no third-party AI service. It sends resource names, alarm and metric values, the change list and your question; never credentials, secrets, environment variable values, log lines or object contents.

It suggests, you decide

Explain never changes anything. A suggested rollback opens the rollback screen, which goes through your usual guardrails; muting an alarm only mutes it on your phone. To dig further, tap Continue in chat: Ask CloudConsole picks up the alarm and answers follow-up questions across your accounts.

What it costs

Explain and the chat are part of Pro. The model runs on your AWS bill at Bedrock’s prices: with Claude Sonnet 5.5, the recommended model, a typical Explain is about a cent, billed by AWS to your account. Setup shows the estimate for the model you choose and a count for the month, and nothing runs in the background. How to set it up.

Questions

Can AI explain why a CloudWatch alarm fired?

It can point to the likely cause when it has the right evidence. CloudConsole collects the alarm, its metric, recent deploys and CloudTrail changes and the resources behind the alarm, and asks Claude on Amazon Bedrock to explain it, citing that evidence.

Does my AWS data leave my account?

It goes from your phone to Amazon Bedrock in your own AWS account, signed with your credentials. CloudConcur never sees it, and Bedrock doesn’t store prompts or train on them by default.

What do I need to set up?

Model access to a Claude model in the Amazon Bedrock console, and the IAM actions bedrock:InvokeModel and bedrock:InvokeModelWithResponseStream for the role you sign in with. Then turn it on in Account → Integrations → Explain & Ask (AI).

How much does it cost?

About a cent per Explain with Claude Sonnet 5.5, billed by AWS to your account, plus CloudConsole Pro. In the demo it works for free on sample data.

Try CloudConsole

On the App Store for iPhone and Google Play for Android. Look around the demo, GitHub and Slack included, before connecting an account.