Voice Commands Reference

Complete list of commands the Voice AI Agent understands

Voice Commands Reference

The Voice AI Agent understands natural language, so you're not limited to exact command syntax. However, this reference provides examples of common commands and explains what each type of request can accomplish. The agent uses OpenAI's function calling capabilities to execute actions, which means it can interpret variations of these commands and still understand your intent.

Content Generation Commands

Generate Images:

• "Create an image of [description]"
• "Generate an image using my [preset name] preset"
• "Make a picture showing [scene description]"
• "I need an image for [purpose] featuring [subjects]"

The agent supports custom descriptions without requiring presets. If you mention a preset that doesn't exist, it will proceed using your description. Image generation uses OpenAI's DALL-E model and typically completes in 10-15 seconds. Each image costs 5 credits.

Generate Copy:

• "Write a LinkedIn post about [topic]"
• "Generate 5 tweets on [subject]"
• "Create email copy for [campaign purpose]"
• "Draft a blog post about [theme]"
• "Give me copy using my [preset name] for [topic]"

Copy generation uses GPT-4 or GPT-5 depending on your plan and the complexity of your request. The agent can create short-form content (tweets, headlines) or long-form content (blog posts, articles). Each piece of copy costs 2 credits.

Generate Videos:

• "Create a 5-second video showing [description]"
• "Generate a product demo video"
• "Make a video with [preset name] about [topic]"
• "I need a video in [aspect ratio] featuring [content]"

Video generation uses OpenAI's Sora 2 model by default (or Google's Veo 3 if specified). Videos typically take 60-120 seconds to generate depending on length and complexity. Each video costs 20 credits. You can specify duration (in seconds) and aspect ratio (16:9, 9:16, 1:1) in your request.

Campaign Management Commands

Create Campaigns:

• "Create a campaign" (AI fills in details from your business context)
• "Start a campaign about [topic] targeting [audience]"
• "Make a campaign with research, blog post, and 3 images"
• "I need a campaign for [objective] with deliverables: [list]"

When you create a campaign without specifying details, the agent uses RAG to query your organization's knowledge base and suggests appropriate campaign names, topics, objectives, and deliverables. It ALWAYS confirms before creating the campaign, showing you the complete plan for approval.

Run Campaigns:

• "Run the [campaign name] campaign"
• "Execute my latest campaign"
• "Start running [campaign name]"

Running a campaign launches the autonomous Campaign Orchestrator agent, which executes all deliverables (research, content creation, creative assets) in an intelligent order without further input from you.

List Campaigns:

• "What campaigns do I have?"
• "Show my campaigns"
• "List all campaigns"
• "What's the status of my campaigns?"

The agent will show your campaigns with their current status (DRAFT, READY, RUNNING, COMPLETED, FAILED).

Research Commands

Immediate Research:

• "Research [topic]"
• "Find information about [subject]"
• "Look up current trends in [industry]"
• "What's happening with [topic] lately?"

Research uses FireCrawl to search the web, scrape relevant articles, and synthesize findings into a structured report. Research costs 10 credits and typically completes in 30-60 seconds depending on query complexity.

Schedule Research:

• "Schedule research on [topic] for [time]"
• "Set up daily research on [subject] at [time]"
• "Research [topic] every [frequency]"
• "In 10 minutes, research [topic]"

The agent understands both absolute times ("tomorrow at 3pm") and relative times ("in 30 minutes"). For recurring research, specify daily, weekly, or monthly with a specific time.

View Research Schedules:

• "Show my research schedules"
• "What research do I have scheduled?"
• "List my active research tasks"

Scheduling Commands

Schedule Content Generation:

• "Schedule an image every Monday at 9am using [preset]"
• "Set up daily copy generation at 2pm about [topic]"
• "Create a video every Friday afternoon"
• "Schedule weekly content generation"

You can schedule images, copy, or videos to be generated automatically at specific times or recurring intervals. Scheduled tasks run in the background and results are saved to your account.

Manage Schedules:

• "Show me my schedules"
• "What do I have scheduled?"
• "List my active schedules"

Email and Communication Commands

Send Emails via Gmail:

• "Send an email to [recipient] about [subject]"
• "Email [contact name] the details about [topic]"
• "Draft and send an email to [address]"

Gmail must be connected in Integrations for email commands to work. The agent will confirm the email content before sending.

Phone Calls:

• "Call my top 10 leads from HubSpot"
• "Make a phone call to [contact]"

Phone call functionality requires the Pro plan or higher and an active HubSpot connection.

Data and Analytics Commands

Query Business Context:

• "What's my company name?"
• "Tell me about my organization"
• "What are our business goals?"
• "What industry am I in?"
• "What's our website traffic this month?"

The agent uses the retrieve_organization_context function to search your synced data and RAG knowledge base for answers.

CRM Queries:

• "How many contacts do I have?"
• "Show me information about [contact name]"
• "Search my CRM for [criteria]"

Requires HubSpot integration to be connected.

List Your Resources:

• "What presets do I have?"
• "Show me my copy presets"
• "List all my video prompts"
• "What image presets are available?"

The agent can enumerate your existing presets and prompts to help you remember what's available.

Utility Commands

Get Current Time:

• "What time is it?"
• "What's the current time?"

Useful when scheduling tasks—the agent uses this internally to calculate relative times.

Help and Information:

• "What can you do?"
• "Help me understand campaigns"
• "How do I [accomplish task]?"

The agent will explain features and guide you through processes.

Best Practices for Voice Commands

Speak Naturally: The agent understands conversational language, so there's no need to use robotic phrasing. "Hey, can you make me an image for Instagram?" works just as well as "Generate image."

Provide Context: When referencing campaigns, presets, or contacts, use names or descriptions. The agent can search for partial matches.

Confirm Before Complex Actions: For multi-step processes or important operations, the agent will always ask for confirmation. Listen to the summary and verify it matches your intent before proceeding.

Use Follow-Up Questions: If the agent's first response isn't quite right, refine your request: "Actually, make it more professional" or "Change the tone to friendly and casual."

Check Results: After content generation, review the output. You can ask the agent to regenerate if you're not satisfied: "Try again with a different style" or "Make another version with more vibrant colors."