This quickstart takes you from a new account to a first workflow that investigates a risk and waits for your approval before it changes anything. Allow about 30 minutes on SaaS, or an hour if you deploy a self-hosted workspace first.
Step 1: Create your account
- Go to autobot.live and enter your details.
- Verify your email through the confirmation link.
- Sign in to open the autobotAI dashboard.
Step 2: Choose where it runs
- SaaS: nothing to deploy. Continue to step 3.
- Self-hosted: deploy a workspace in your AWS account with CloudFormation. It typically takes 30 to 45 minutes. Follow Deploy a self-hosted workspace, then come back here.
Deployment and models compares the two.
Step 3: Connect a model
- Open Integrations and click + Add Integration.
- Choose your model provider, such as Amazon Bedrock, OpenAI or OpenRouter, and enter its credentials.
- Open User Settings from your profile icon, go to Settings, and select the integration under AI Settings so AI Builder, Optimus and AI nodes use it by default. See Connect your AI models.
Step 4: Connect your first tools
Add the integrations your first use case needs, from the same Integrations page:
- The system the risk lives in, for example AWS, Microsoft Entra ID or GitHub.
- Where approvers work: Slack, Microsoft Teams or email.
- Any tool that adds context, such as VirusTotal or your SIEM.
After saving each one, run its test to confirm the connection.
Step 5: Build your first workflow
Choose the way that suits you. In the app, workflows are listed under Action Resources > Bots.
| Way to build | Best when | Guide |
|---|---|---|
| Start from the library | A ready-made template covers your use case | Start from the library |
| Describe it to AI Builder | You know the outcome and want a first draft on the canvas | Build with AI Builder and Optimus |
| Ask Optimus | You want to try the job ad hoc first, or it needs a harness of several agents | Investigate with Optimus |
| Build node by node | You want full control of every step | Build a bot node by node |
For example, ask AI Builder: "Every day, find IAM access keys older than 90 days, look up each key's owner, and ask the owner's manager in Slack to approve deactivating it."
Not sure whether the job needs a workflow or a harness? See Choosing a workflow or a harness.
Step 6: Add an approval before any change
Before a step that changes anything, add an approval node and name the approver. The run will pause there and send the request to Slack, Microsoft Teams or email, with the evidence the workflow gathered.
Start new workflows this way. You can give an action more autonomy later, once its record shows it's safe. See Autonomy and approvals.
Step 7: Run it and review
- Run the workflow. Scheduled workflows can also be run manually from the Bots page.
- Watch it progress in Bot executions: its status, what triggered it, each step it took and the results.
- Approve or reject the request when it arrives. Approvers can also decide from the Approvals page.
- Check the audit trail for the full record of who did what, when and why.
Next steps
- How autobotAI works: follow a risk from detected to closed.
- Agents, workflows and harnesses: grow from one workflow to a harness of agents.
- Architecture and security: what runs where, and what stays in your account.
- Integrations: connect the rest of your stack.
- Role-based access control: invite your team with the right permissions.