Chats

AI Chat Features

Learn about AI Chat Features in WaymakerOS.

ArtifactsPromptsMentionsModelsRegenerate

Beyond basic conversation, Chats includes advanced AI capabilities for content generation, data analysis, and external service integration.

Artifacts

Artifacts are structured content generated by the AI — documents, code, spreadsheets, diagrams, and more. When the AI creates an artifact, it appears as a clickable card in the conversation.

Artifact types

TypeDescription
DocumentRich text content with headings, lists, tables
CodeSyntax-highlighted code in any language
SheetTabular data with headers and rows
TableStructured JSON data in grid format
DiagramFlowcharts, sequence diagrams, entity-relationship diagrams (Mermaid syntax)
PresentationSlide deck content
FormForm definitions with fields and validation
TaskboardKanban board structure

Saving artifacts

  1. Click the artifact card in the conversation
  2. The preview panel opens on the right
  3. Review the content in native format
  4. Click Save
  5. Choose a workspace and project
  6. Give it a title
  7. The artifact is saved as a new Commander tool (document, sheet, etc.)

Tip: You can copy artifact content to clipboard without saving by clicking the copy icon in the preview panel header.

Prompt Library

The Prompt Library helps you get consistent, high-quality AI responses by providing pre-built context for your messages.

Waymaker prompts

Built-in prompts based on proven business frameworks. These cover strategy, planning, analysis, and leadership topics.

Partner exclusive prompts

Prompts from Waymaker partners that give specialised guidance for specific industries and use cases. They appear under Partner Exclusive for every organisation, whatever your plan. A partner prompt that needs a partner licence shows a crown icon and can't be added to a chat yet.

Custom prompts

Create your own reusable prompts:

  1. Click New in the Prompt Library header
  2. Enter a title and description
  3. Write the prompt text
  4. Choose a category
  5. Save

Your custom prompts appear under My Custom Prompts and are available to you across all chats.

Using prompts

  1. Open the Prompt Library sidebar
  2. Search or browse for a prompt
  3. Click the + button to add it as context
  4. The prompt appears as a badge in the input bar
  5. Type your message — the AI will consider the prompt when responding
  6. Click X on the badge to remove it before sending

Mentions

Reference Commander resources in your messages using the @ symbol:

  1. Type @ in the input bar
  2. A searchable dropdown appears
  3. Select a task, document, sheet, or other resource
  4. The mention is inserted into your message

The AI can read and reference mentioned resources in its response, giving you contextual answers about your actual work.

Connected services

If your organisation has external services connected (such as Google Analytics or Google Search Console), you can query them directly from Chats. The AI detects when your question relates to a connected service and fetches live data.

For example:

  • "What were our top pages by traffic last month?"
  • "Show me our search console performance for the past week"

Results appear formatted in the conversation with source attribution.

Real-time collaboration

Multiple people can participate in a shared chat:

  • Presence indicators show who is currently viewing the chat
  • Typing indicators show when someone is composing a message
  • Live updates — new messages appear instantly for all participants

Choosing a model

The picker beside the message box sets which AI answers in this chat. It has two parts.

Mode lets One choose for you:

ModeWhat answers
AutoFast for most messages; Smart for long messages and images
FastClaude Haiku: quickest, usually under 2 credits a message
SmartClaude Sonnet: deeper answers, about 20 times the credits of Fast

Choose a model picks one model for every message in the chat. Each option shows who makes it and roughly how many credits a typical message costs. The costs are approximate: a long conversation, a web search (10 credits) or reading pages costs more.

GroupModelMade byTypical message
ClaudeClaude HaikuAnthropic~1 credit
ClaudeClaude SonnetAnthropic~14 credits
ClaudeClaude OpusAnthropic~28 credits
ClaudeClaude FableAnthropic~70 credits
Other frontier modelsGPTOpenAI~14 credits
Other frontier modelsGPT FastOpenAI~1 credit
Other frontier modelsGrokxAI~20 credits
Low-cost open modelsDeepSeek FlashDeepSeek~2 credits
Low-cost open modelsGemma 4Google~1 credit

Some models carry a note in the picker:

  • Claude Opus needs 1,300 credits free to start a message, and Claude Fable needs 3,200. Credits are held at the start and you are charged only what the message actually used.
  • GPT can't use MyVault yet, and it answers without web search.
  • Claude Fable is offered only in a chat in a workspace marked as not handling health information.

When a model isn't available

A model you can't use here is shown greyed out, with the reason:

  • This workspace handles health information. A workspace that handles health information always uses Claude Sonnet, and the picker shows only that.
  • Needs a chat in a workspace. The models outside Claude Haiku, Sonnet and Opus need a chat that belongs to a workspace.
  • Not switched on for your organisation yet.
  • Only the owner or an editor can change the model. Anyone else in the chat sees the model that was chosen.

If the model you chose can't answer a message, another model answers it and a line under the reply says so, for example "Answered by Claude Sonnet: Grok isn't available for this message." A model is never swapped silently.

Retired models

Two models that were offered are being retired by their provider and are no longer in the picker:

Retired modelReplaced byStops working
DeepSeek V4 ProDeepSeek Flash20 November 2026
Kimi K2.6Gemma 420 November 2026

A chat that had one of them chosen keeps working: its messages are answered by the replacement, which costs no more, and the reply says why. You can't choose a retired model again.

Regenerating a reply and editing your message

You can redo the latest exchange in a chat in two ways:

  • Regenerate, under the latest reply, asks again for a new reply to the same message.
  • Edit, under your latest message, opens it in an editor. Change the text and click Send (or press Enter; Shift+Enter adds a new line). Cancel or Escape closes the editor without sending.

What to know:

  • Only the latest message, and only your own. The buttons appear only on the last message in the chat, and only for the person who sent it. If someone else has sent a message since, there is nothing to regenerate or edit. A person who can only view the chat never sees them.
  • It is charged like any message. A regenerated or edited message uses the chat's model and costs the same credits as sending it fresh.
  • Nothing is deleted. The original message and its reply stay in the chat above the new one, dimmed and labelled Earlier version. The AI no longer reads them: the conversation carries on from the new version. Other people in the chat see the earlier version dimmed after they reload the chat.
  • Guests don't see earlier versions. Someone reading the chat through a guest link sees only the current version.
  • Images aren't sent again. Mentions in the original message are sent again with it; images you uploaded with it are not.
  • A message you stopped before it was answered can't be regenerated or edited. Send a new message instead.

Token usage

AI responses consume credits from your organisation's allocation. The chat interface shows your remaining credit balance. Credit usage varies by:

  • Message length and complexity
  • The mode or model chosen (Choosing a model)
  • Whether external service queries are involved

Next steps