> ## Documentation Index
> Fetch the complete documentation index at: https://notiky-mintlify-b5788c9c.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Conversations

> Ask, Think, Plan, and Do — operator sessions for product work, not generic chat.

Conversations are how you **think through product work** inside Notiky. Pick a mode that matches how much the AI should read vs act — same idea as giving a colleague a brief vs asking them to ship.

<Frame>
  <img src="https://mintcdn.com/notiky-mintlify-b5788c9c/kOTGsdF2jDE8PXYg/images/guides/conversations-modes.png?fit=max&auto=format&n=kOTGsdF2jDE8PXYg&q=85&s=e277433a47a0e77e67304481c2507ed0" alt="Conversation composer with Plan mode selected for shaping work" width="1176" height="828" data-path="images/guides/conversations-modes.png" />
</Frame>

***

## The four modes

| Mode      | Best for          | What it can do                                            |
| --------- | ----------------- | --------------------------------------------------------- |
| **Ask**   | Quick questions   | Read workspace context — no changes                       |
| **Think** | Exploring options | Read + capture decisions and notes                        |
| **Plan**  | Shaping work      | Draft tasks and artifacts — no live mutations             |
| **Do**    | Shipping          | Full operator tools — create/update tasks, publish drafts |

Switch modes with the pills in the conversation composer — **Ask**, **Think**, **Plan**, **Do**.

***

## When to use each mode

### Ask

You want an answer from your workspace: “What did we decide about pricing?” or “Summarize open tasks for this project.”

### Think

You are working through a decision. Notiky can challenge assumptions and capture structured notes — aligned with the landing promise of an AI that **argues back**, not agrees blindly.

### Plan

You are ready to break work into tasks or drafts but not execute yet. Good for PMs shaping a sprint before assigning agents.

### Do

You want the agent to take action: create tasks, update the board, publish artifact drafts. Use when you trust the direction and want execution.

***

## Typical PM flow

<Steps>
  <Step title="Think through the decision">
    Open a conversation in **Think** — explore options and capture rationale.
  </Step>

  <Step title="Plan the work">
    Switch to **Plan** — draft tasks with enough context for a developer or agent.
  </Step>

  <Step title="Assign to an agent">
    Move tasks to the board and assign an [agent](/agents) on a connected runtime.
  </Step>

  <Step title="Track on the board">
    Follow progress in [Tasks](/tasks) like any other teammate.
  </Step>
</Steps>

See the full recipe: [Decision → task → agent](/example-workflows/decision-to-task).

***

## Related

<CardGroup cols={2}>
  <Card title="Tasks" href="/tasks" />

  <Card title="Agents" href="/agents" />

  <Card title="Capabilities" href="/capabilities" />

  <Card title="Key concepts" href="/concepts/key-concepts" />
</CardGroup>