Charms are not chatbots. They don't live in a sidebar waiting for questions. They live 

 your document — reading what you've written, seeing your reference materials, and producing work that appears right alongside yours. They're participants in the document, not spectators.

When you tag a Charm in your document, here's what happens:

This is fundamentally different from a chat interface. In ChatGPT, you get a response in a conversation thread. In CharmIQ, the Charm's output becomes part of the document itself. It's your draft, your analysis, your slide deck — not a chat log you have to copy-paste from.

Every Charm has three things you control:

Different models think differently. CharmIQ gives you access to all of them:

You're not locked into one. You can have different Charms using different models in the same document. A Claude Charm for brainstorming, a Gemini Charm for building slides. Pick the right tool for the job.

A Charm with a specific persona produces dramatically sharper output than a generic one. Instead of "marketing expert," try "Mark Ritson — contrarian marketing strategist who insists on evidence-based approaches." The more specific, the better.

This is CharmIQ's real differentiator. You control exactly what information a Charm has access to:

 — Attach PDFs, documents, images, or other CharmIQ workspaces. The Charm reads them as context when generating its response. Drop in a competitor's annual report, a research paper, your company's brand guidelines — whatever the Charm needs to do its job well.

 — Control which parts of a long document the Charm can see. Working on section 5? Set a breakpoint so the Charm focuses on sections 4 and 5, not the exploratory mess from sections 1–3.

 — Everything you've written, every annotation, every previous Charm output above the current prompt. The Charm reads it all (within the breakpoint scope). This means your document is a 

Tag a Charm and ask it to produce something new:

"Write a competitive analysis of our three main competitors, focusing on pricing strategy and market positioning."

The Charm writes the analysis directly into your document.

 (not the main document) when you want a Charm to edit something that's already there. Tag the Charm in a comment and ask:

"Rewrite the executive summary to be half the length and twice as direct."

The Charm reaches back into the document and modifies the existing text.

Attach a document as Reference Material, then ask the Charm to work with it:

"Based on the attached research report, identify the three findings most relevant to our product roadmap and explain why."

Tag a Charm (Gemini works best for this) and ask for slides:

"Create a modern elegant slide from the above information."

A workspace isn't a one-on-one conversation. It's a room you fill with the right people for the job.

: Tag different Charms at different points in your workflow. One generates a draft, another critiques it, a third rewrites based on the critique. Each Charm sees everything that came before it — including the other Charms' contributions.

: After completing a major piece of work, switch to a more powerful model (like Claude Opus) and ask one question: 

"Is there anything I haven't considered?"

 Slower, more expensive — but the depth on a single final-pass review is often worth it.

: Not sure about a Charm's output? Tag a different Charm and ask it to evaluate. One Charm wrote the research brief; another Charm with domain expertise reviews it for blind spots.

Charms use AI credits every time they generate content. Different models cost different amounts, and longer outputs use more credits.

You don't need credits to learn CharmIQ. Start with the workspace. Add AI when you're ready.

 — not by you personally. If you belong to multiple Organizations (a personal account and a Company, for example), make sure you're in the right one before running credit-intensive work.

 — How Organizations work, how billing is tied to them, and how to switch between them.

 — See Charms in action in a 10-minute walkthrough.

 — Haven't explored the workspace yet? Start there.

AI agents that read your context, think alongside you, and write directly into your workspace.

Working with Charms

What a Charm Actually Does

Configuring a Charm

The AI Model

The Persona

What They See (Context Control)

The Core Interaction Patterns

Write Fresh Content

Revise Existing Content

Analyze Reference Material

Generate Visual Output

Multiple Charms, One Document

Credits and Costs

Next Steps