Edit, Plan, and Ask

SpaceChat has three modes that control how much the AI can act. Each one is a guarantee.

Ask#

Discussion only. The AI can read and reason. It cannot mutate state.

Use Ask when:

• You're exploring options
• You want a second opinion before deciding
• You want to understand a doc, an email, a project
• You'd be unhappy if the AI did something you didn't expect

Guarantee: nothing happens. No entities created, no emails sent, no side effects.

Plan#

The AI researches and produces a structured action plan. It stops there.

Use Plan when:

• The task is multi-step and you want to inspect the plan first
• The work is risky enough that you want explicit review before execution
• You're going to do something but want to make sure the AI agrees with the approach

Guarantee: the AI writes a plan, but doesn't act on it. After review, switch to Edit and ask the AI to execute.

Edit#

The AI does the work — calls tools, writes entities, sends messages, takes action.

Use Edit when:

• The task is routine
• The right answer is well-defined
• You've already approved a plan

Guarantee: the AI acts. You see what it did in the chat thread and in Activity.

How to choose, fast#

• Still exploring? Ask.
• Want to see the plan first? Plan.
• Ready to ship? Edit.

Switching mid-conversation#

You can flip between modes inside a single thread without losing context. A common rhythm:

1. Ask — "what's the state of the launch doc?"
2. Plan — "what would I need to update to ship Friday?"
3. Edit — "do everything in that plan."

The AI keeps the conversation across mode switches. You're just changing the rules of engagement.

When in doubt#

Use Plan. It's cheap to read; mistakes in Edit can be expensive.

The three-mode pattern is how teams stay intentional about AI autonomy without giving up the speed AI provides.