Tool Pills in the Prompt Bar
Tool pills are toggle buttons that sit in the prompt bar below the text input. Click a pill to activate it (highlighted = on); click again to deactivate. The tool state applies to the message you send next.| Tool | What It Does | When to Use It | Plan |
|---|---|---|---|
| Web Search | Searches the web in real time and cites sources | Current events, fact-checking, recent news, anything the model may not know | Free: 1/day · Pro+: unlimited |
| Deep Research | Runs a comprehensive multi-step research report across many sources | Industry analysis, competitive intelligence, academic research, thorough topic overviews | Free: 1/day · Pro+: unlimited |
| Canvas Mode | Opens an interactive editor alongside the chat | Writing documents, editing code, building spreadsheets, creating diagrams, designing slides | All plans |
| Agent Mode | Takes autonomous actions across connected apps in a multi-step loop | Multi-step workflows, cross-app automation, complex tasks requiring tool calls in sequence | Pro+ (limited on Free) |
| Voice | Starts a real-time WebRTC voice conversation | Hands-free use, brainstorming out loud, accessibility, conversational ideation | All plans |
How to Enable or Disable a Tool
- Locate the pill buttons in the prompt bar
- Click a pill to enable it — it will appear highlighted or outlined in the active color
- Click the same pill again to disable it
- Send your message — the active tools apply to that send
Web Search
Web Search gives the model access to live internet results. When enabled, ZeroTwo queries the web before generating its response and includes source citations inline.Focus Modes
Web Search includes 8 focus modes that narrow where ZeroTwo looks:| Focus Mode | What It Searches |
|---|---|
| Auto | General web search (default) |
| News | Recent news articles |
| Research | Academic and scholarly sources |
| Company | Business and company information |
| Publicly accessible PDF documents | |
| GitHub | GitHub repositories and code |
| Personal | Your connected personal data (via connectors) |
| LinkedIn profiles and company pages |
Deep Research
Deep Research is designed for thorough, multi-step research tasks. Instead of a single search, ZeroTwo breaks the question into sub-queries, searches multiple sources across the web, synthesizes the results, and produces a structured report with full citations. Use Deep Research when you need:- A comprehensive overview of a topic or industry
- Comparison of options with detailed pros and cons
- Literature reviews or competitive analysis
- Research that would normally take hours of manual effort
Canvas Mode
Canvas Mode opens an interactive editing surface alongside the chat. Instead of returning plain text, the AI writes directly into a live editor that you can continue editing, formatting, and exporting. ZeroTwo automatically selects the most appropriate canvas editor based on your request, or you can switch editors manually:| Canvas Type | Description |
|---|---|
| Rich Text | Full document editor with headings, lists, tables, and inline formatting |
| Code | Syntax-highlighted code editor with language detection |
| Spreadsheet | Grid-based spreadsheet editor for tabular data |
| Mermaid Diagram | Diagram editor that renders flowcharts, sequence diagrams, Gantt charts, and more |
| HTML Preview | Live HTML editor with side-by-side rendered preview |
| SVG | Vector graphic editor |
| PDF Viewer | Display and annotate uploaded PDFs |
| Slides | View and navigate presentation slides |
Agent Mode
Agent Mode enables ZeroTwo to take multiple autonomous steps to complete a task. Instead of a single response, the AI reasons about what actions are needed, calls tools in sequence, observes results, and iterates until the task is complete. Agent Mode is most valuable for:- Complex multi-step workflows that span multiple tools
- Cross-app automation (for example: search the web, draft a document, save it to Google Drive)
- Code generation tasks that involve writing, running, and testing steps
- Research-to-output pipelines
Voice
Voice activates a real-time WebRTC audio conversation with the AI. You speak and hear responses in a natural voice — the conversation is also transcribed and saved to your chat history. ZeroTwo offers 10 voice options:| Voice | Character |
|---|---|
| Alloy | Neutral, clear |
| Ash | Warm, conversational |
| Ballad | Expressive, lyrical |
| Cedar | Calm, measured |
| Coral | Bright, friendly |
| Echo | Deep, resonant |
| Marin | Crisp, professional |
| Sage | Thoughtful, deliberate |
| Shimmer | Light, upbeat |
| Verse | Articulate, balanced |
File Uploads
File uploads are not a pill — use the paperclip icon in the prompt bar to attach files to any message. You can also drag and drop files onto the chat area.- Free plan: 1 file per request
- Pro and above: Up to 10 files per request
Connectors (@ Mentions)
Connect your external apps in Settings → Connectors. Once connected, invoke any connector by typing@ in the prompt bar and selecting from the dropdown.
Examples:
Studio Tools
In addition to the prompt bar tools, ZeroTwo has three dedicated Studio pages for media generation:Image Studio
Generate AI images at /studio/images with controls for style, aspect ratio, and model selection.
Video Studio
Generate and edit video content at /studio/video.
Audio Studio
Generate voiceovers and audio at /studio/audio.
Files Library
Browse and reuse all generated and uploaded files at /files.