Quickstart Guide

Welcome to Cesteral Intelligence. This guide walks you through the actual first-run experience so you know what to expect.

Create Your Account

1. Visit the homepage and click Get Started
2. Sign up with your email or use single sign-on
3. Verify your email address
4. You'll be redirected to your personal workspace

Explore Home

Your workspace opens to Home (/workspace) — an overview of what's happening and what needs you:

• Governance KPIs — action approval rates, confidence trends, policy pass rates
• Recent runs — latest assistant conversations and governed actions
• Daily briefing — an AI-generated summary of what happened and what needs attention

Meet the Assistant

Open Chat (under Operate in the sidebar) to talk to the governed assistant — the conversational way to get work done. The assistant can:

• Answer questions about your campaigns and data
• Propose actions (ad launches, budget changes, audience updates)
• Execute approved actions through governed automations
• Pull context from your knowledge stack automatically

Start a conversation by typing a message. The assistant has access to your project memory, knowledge records, and connected documents from the moment you begin. Every action it proposes is evaluated by governance before anything runs.

Review Pending Approvals

Open Queue in the sidebar, under the Governance group. The queue shows pending actions waiting for your review:

• Each action includes a structured manifest showing exactly what will change before execution
• A confidence score (0–1) tells you how certain the AI is about the action
• Actions below your configured threshold, or that trip a blocking rule, are held here for manual review

Approve or reject each action from this view. Approved actions move to execution; rejected actions are recorded in the audit trail. Bundled, multi-step campaign changes appear under Changes, in the same group.

Check the Approval History

Open Audit log, also under Governance, to see completed governance decisions:

• Every policy evaluation result with the rule that fired
• Every approval and rejection with timestamps
• Deep-link to individual executions and decisions via the URL

Configure Your First Guardrail

Open Policies in the sidebar, click Create policy, and choose Block unsafe work to reach Blocking rules. This is where you define the JSONLogic policies that evaluate every AI action:

• Set spending limits, targeting constraints, brand safety rules
• Configure severity (warning vs. error/block)
• Scope policies to specific action types

See Guardrails for a full walkthrough.

Next Steps

• Learn the platform model in Key Concepts — how the workspace sidebar is organized, and how the knowledge stack works
• Set up your knowledge stack in Knowledge
• Understand Guardrails and how policies evaluate every action
• Explore Skills to encode your team's standard operating procedures