Getting started with Craig

This guide walks you through everything you need to get Craig up and running — from understanding what it is, to building your first Prompt and Agent. Follow the steps in order and you'll have a solid foundation to build on.

If you're already familiar with Craig and looking for something specific, use the category navigation to go directly to the article you need.

Step 1 - Understand what Craig is and whether it's right for you

Before configuring anything, it's worth taking a few minutes to understand what Craig does, how it handles your data, and whether it's a good fit for your team. Craig is entirely optional — your Connect Rocket account works exactly the same whether you use it or not.

• What is Craig? — An overview of Craig's features and where they fit within Teams and Community.
• Craig: AI Safety Overview — What Craig can and can't do, and the guardrails built into how it operates.
• Do I have to use Craig? — Short answer: no. This article explains your options.

Step 2 - Understand how Craig handles your data

If you're an administrator or you've been asked to evaluate Craig for your organisation, this article covers what data Craig can access, what it never touches, and the infrastructure controls in place.

• Craig: Data Handling and Privacy — Intended for administrators and technical reviewers. Covers data flows, AWS Bedrock posture, tenant isolation, and the audit trail.

Step 3 - Set up your organization's context

Contexts are reference notes that help Craig understand your organization — local place names, equipment identifiers, terminology your team uses day-to-day. Setting these up early means Craig's responses will be more accurate and relevant from the start.

• What are Contexts? — What contexts are and how Craig uses them.
• Configuring Contexts — How to add and manage context entries. Administrator access required.

Step 4 - Explore available data sources

Craig retrieves data from a curated library of public and government sources called Tools. Before building Prompts or Agents, it's useful to know what data is available to you.

• What are Tools? — An explanation of what Tools are and how Craig uses them.
• How can I view available Craig Tools? — How to browse the full Tools library from within the web app. Administrator access required.

Step 5 - Build and run your first Prompt

Prompts are pre-configured instructions that direct Craig to fetch current data and deliver it as a formatted briefing. Start here if you want to run on-demand situational awareness briefings or integrate data into your notification workflow.

• What are Prompts? — What Prompts are, how they work, and when to use them.
• Configuring Prompts — Step-by-step instructions for building a Prompt. Administrator access required.
• Enable Craig Prompts on the Dashboard — How to make Prompts available to your team from the Dashboard.
• Using Prompts from the Dashboard — How to run a Prompt and review the results.
• Using Prompts in Workflows — How to trigger a Prompt automatically as part of a Workflow.

Not sure what to build? Examples of Prompts shows how teams like yours are using Prompts across a range of operational scenarios.

Step 6 - Set up your first Agent

Agents monitor conditions you define and notify the right people when a threshold is met — automatically, without anyone having to check manually. Start here once you're comfortable with Prompts.

• What are Agents? — What Agents are and how they work.
• Configuring Agents — Step-by-step instructions for building an Agent. Administrator access required.
• Writing Good Conditions for Craig Agents — How to write conditions that evaluate reliably and fire when they should.

Not sure what to build? Examples of Agents shows how teams like yours are using Agents to monitor conditions across a range of operational scenarios.

Step 7 - Go further

Once you have your first Prompts and Agents running, these articles will help you get more out of them.

• Fast and Thinking: How Craig Chooses the Right Model — Understand how Craig assigns a processing model to each Prompt and Agent, what it means for cost, and how to write efficiently.
• Common AI Mistakes to Avoid When Writing Craig Prompts — The most common prompting mistakes and how to fix them. Essential reading before building anything complex.
• Writing Good Conditions for Craig Agents — Worth revisiting once you've built your first Agent and want to refine it.

A note on credits

Craig uses credits each time a Prompt is run or an Agent checks a condition. The credit cost for each Prompt and Agent is displayed in the UI alongside the model icon — so you always know what a run will cost before it happens. For more detail on how model assignment affects credit usage, see Fast and Thinking: How Craig Chooses the Right Model.