Skip to main content
This quickstart takes you from sign-in to your first AI investigation: CloudThinker works out a root cause and proposes a fix. Allow about 10 minutes — and no cloud credentials if you start with the demo.
What you’ll need: a Google or GitHub account, or any work email. Your own AWS, Azure, Google Cloud, or Kubernetes access is optional — the read-only demo works without it.

​
The 30-second syntax

Talk to agents in plain language. The full pattern is:
  • @agent (optional) — leave it out and CloudThinker answers, or mention one of your workspace’s custom agents.
  • #connection (optional) — the connection to use, such as your AWS account or a Kubernetes cluster.
  • / (optional) — run a skill or a command.
  • Your request — what you need, in your own words. Add any context inline.
See Commands for the full list.
1

Sign in

Go to app.cloudthinker.io and sign in with Google, GitHub, or SSO. To use email instead, enter your address, click Continue with email, and type the 6-digit code CloudThinker sends you.The first time you sign in, two short questions follow: Tell us about you (your role) and, for the organization owner, How can I help you? Answer them or click Skip — choose Just exploring to follow this guide.Success state: you land on Home, with a chat box and the Needs you, Modules, and Automation today cards. The sidebar shows New chat, Artifacts, Automations, and Investigation.
A new organization starts on a 14-day Team trial with no card required. See Pricing for plan limits.
2

Connect the demo or your own cloud

  1. Click Try a demo environment in the sidebar, or go to Settings → Connections.
  2. Pick a service marked Demo available — AWS, Google Cloud, Azure, or Kubernetes, whichever is offered.
  3. Click Connect, then Use demo. Setup takes about a minute.
The demo environment is read-only and shared with other people trying it. Demo chats use credits like any other chat.
Success state: the connection shows Connected. If it shows Connection error, see Troubleshooting.
3

Ask your first question

With the demo, CloudThinker opens a chat with an editable starter prompt, such as Give me an overview of this AWS account. Send it as is.With your own cloud, click New chat and send:
Swap in your provider or cluster name.Success state: CloudThinker answers with services, regions, and resources you recognize from the account — not generic best practices.
4

Run your first investigation

  1. Click Investigation in the sidebar. If Resolve asks you to connect your first alert source, click Skip for now — the next step covers it.
  2. Click Log Incident.
  3. In What’s happening?, describe a real problem — for example, something the previous answer flagged. Or pick a Quick Start Template such as Service Degradation to pre-fill the form.
  4. Choose a Severity, add any detail under Context, and click Log Incident.
CloudThinker starts root-cause analysis right away. A manually logged incident needs no alert source.Success state: the incident opens in the Investigation view under AI is handling. When the investigation finishes, it shows a root cause with evidence, a timeline, and remediation suggestions for you to approve — see Investigation and RCA for how to read it.
5

Investigate every incident automatically (optional)

Go to Resolve and click Connect a source. Choose a webhook (Datadog, PagerDuty, Grafana, Opsgenie, or custom alerts), cloud polling (AWS, Azure Monitor, Google Cloud Monitoring), or Slack or Microsoft Teams.On Duty — automatic investigation — is on by default, so every incident Pulse routes from that source gets investigated without anyone clicking Investigate. The full setup is in Connect Pulse sources.Success state: an event sent from your tool appears in Pulse as a Signal.

​
You’re done when…

  • A connection — demo or your own — shows Connected
  • An agent answered with resources you recognize from that environment
  • Your logged incident shows its investigation result, and you can trace it back through the evidence
The Getting started checklist tracks these first steps and adds credits as you complete them. From your next sign-in, Home is your starting point: Needs you lists anything waiting on your approval, and a banner appears for any live incident.

​
Troubleshooting

Check: your spam or promotions folder, and that the address on the code screen is spelled correctly.Fix: click Resend code, or sign in with Google or GitHub instead.
Likely cause: a demo is offered only while its demo target is available.Fix: click Show all connections and check the other providers, try again later, or connect your own cloud instead.
Likely cause: the credential is wrong or lacks permissions — for AWS, usually an External ID mismatch in the role’s trust policy.Fix: correct the credential in your cloud provider, then click Retry on the connection. Provider-specific checks are in each connection guide, such as AWS troubleshooting.
Likely cause: the connection lives in a different workspace from the chat you’re in.Fix: check the workspace selector at the top of the sidebar, switch to the workspace where you added the connection, and ask again.
Likely cause: the available evidence didn’t support a root cause — common for an incident that never happened in the connected environment.Fix: log an incident about something real in that environment, add detail under Context, or connect the cloud or monitoring tool that holds the evidence.

​
Next steps

Read an investigation

Follow the root cause, evidence, and remediation queue, and settle each suggestion

First prompts by role

Find your role, copy a prompt, and get a real result from your own environment

How CloudThinker fits together

See the whole model — workspaces, connections, agents, guardrails, and modules — on one page

Invite your team

Add teammates with Owner, Admin, Developer, or Viewer roles and assign workspace access