Workjet
Back to Workjet

Chat

Interactive chat is the primary interface for conversational AI in Workjet. The chat panel is docked at the bottom of every tab, providing instant access to AI from anywhere in the workspace. Every message flows through the AI gateway for DLP scanning, cost tracking, and audit logging.

Sending Messages

Type your message in the chat input and press Enter to send. Use Shift + Enter for multi-line input. Responses stream in real time with token counts displayed as they arrive.

Document Drag-and-Drop

Drag a file onto the chat panel to include it as context for your conversation. The AI reads the document and can answer questions about its contents, summarize it, extract data, or transform it.

This is useful for tasks like:

  • Summarizing a lengthy PDF report
  • Reviewing and suggesting improvements to a code file
  • Extracting structured data from a CSV
  • Analyzing meeting notes in a markdown file

File Attachment

Click the attachment icon in the chat input to browse and attach files. Supported file types include:

Category Formats
Documents .pdf, .docx, .txt, .rtf
Code .js, .ts, .py, .go, .rs, .java, .cpp, .html, .css, and more
Data .csv, .json, .xml, .yaml
Markup .md, .mdx, .tex

Tip: You can attach multiple files in a single message to compare documents, merge data, or provide broader context for your question.

Model Switching

Use the model selector dropdown in the chat header to switch between available AI models mid-conversation. Available models depend on your organization's routing configuration and may include:

  • Anthropic: Claude Opus, Claude Sonnet, Claude Haiku
  • OpenAI: GPT-4o, GPT-4o Mini
  • Google: Gemini Pro, Gemini Flash
  • Open source (via Ollama): Llama, Qwen, DeepSeek, Gemma

Switching models does not reset the conversation — the full message history is sent to the new model so it has complete context.

Streaming Responses

Responses stream token-by-token as the model generates them. You'll see:

  • Text appearing in real time as the model writes
  • A running token count in the message footer
  • A stop button to cancel generation mid-stream

Document Context

When you attach or drag a document, it becomes part of the conversation context. The AI can reference the document across multiple turns. Document context persists for the entire chat session, so follow-up questions can reference previously attached files.

Chat Commands

Special commands can be typed directly in the chat input:

Command Description
/clear Clear the current conversation history
/model [name] Switch to a specific model
/skill [name] Invoke a skill by name
/export Export the conversation as markdown

Skill-Enhanced Chat

You can invoke skills directly from the chat. When a skill is activated, its system prompt and tool configuration are applied to the conversation, giving you a specialized AI assistant for the duration of that interaction.

Next Steps