Skip to content

AI Cold DM

Generates and sends AI-powered personalized Direct Messages from a user’s profile page.

  1. Takes a screenshot of the profile (for logging)
  2. Extracts username, full name, and bio from the profile page
  3. Clicks the Message button on the profile
  4. Waits for the DM screen to load
  5. Sends profile context to an AI provider for personalized message generation
  6. Types the message with human-like simulation
  7. Sends the message
  8. Verifies delivery (input emptied + send button disappeared)
  9. Optionally returns to the profile page
  10. Logs the action for tracking
ProviderModels Available
OpenAIGPT-4o, GPT-4o Mini, GPT-4 Turbo, GPT-4 Vision Preview
Claude (Anthropic)Claude 3.5 Sonnet, Claude 3 Opus, Claude 3 Sonnet
DeepSeekDeepSeek Chat, DeepSeek Reasoner
Gemini (Google)Gemini 1.5 Pro, Gemini 1.5 Flash, Gemini Pro Vision
GroqLlama 3.1 70B, Mixtral 8x7B
CustomAny OpenAI-compatible endpoint
OptionTypeDescription
AccountSelectAccount for logging and screenshot storage
ProviderSelectAI provider to use
API KeyPasswordYour API key (stored securely in node)
ModelSelectAI model (options depend on provider)
Custom Model NameTextFor custom model names (when “Custom” selected)
Custom Endpoint URLTextFor custom providers (OpenAI-compatible format)
PromptTextareaInstructions for AI message generation (up to 8000 chars)
Min LengthNumberMinimum message length in words (default: 3)
Max LengthNumberMaximum message length in words (default: 500)
BlacklistTextareaWords to avoid, one per line
Enable Typing SimulationCheckboxType like a human (recommended)
Typing PresetSelectFast, Normal, Slow, Human, or Custom
Custom Typing ConfigObjectMin/max delay, chars per word (for Custom preset)
Go Back After SendingCheckboxReturn to profile page after sending (default: on)
Wait AfterNumberSeconds to wait after sending (default: 1.0)
PresetSpeedRealism
FastQuickLow
NormalModerateMedium
SlowSlowHigh
HumanVariableVery High
CustomConfigurableConfigurable
FeatureDescription
Profile ContextAI uses username, full name, and bio to personalize the DM
Text-Only GenerationNo screenshot sent to AI — profile data extracted via XML for speed
Multi-Provider Support6 AI providers + custom endpoints
Human-Like TypingConfigurable typing simulation with natural delays
Auto-RetryAutomatic retry on API failures (3 attempts, exponential backoff)
Blacklist FilteringAvoid specific words in generated messages
Length ValidationEnsures messages meet min/max word requirements
Send VerificationConfirms message was sent (input emptied + send button gone)
Go BackOptionally returns to profile page after sending
Action LoggingLogs all DMs to JSONL file for tracking
Screenshot StorageSaves profile screenshots for reference
DataLocation
Profile ScreenshotsAccounts/{account}/dms/cold/
DM LogAccounts/{account}/dms/dm_log.jsonl
{
"type": "cold_dm",
"target_username": "john_doe",
"dm_text": "Generated DM text",
"provider": "openai",
"model": "gpt-4o-mini",
"success": true,
"account": "your_account",
"screenshot": "/path/to/screenshot.jpg",
"timestamp": "2024-01-15T10:30:00Z"
}

Returns:

  • success: Whether the DM was sent
  • provider: AI provider used
  • model: Model used
  • dm_text: The generated DM text
  • target_username: Target user’s username
  • screenshot: Path to saved profile screenshot
  • duration: Execution time in seconds
  • API key for chosen provider
  • Internet connection
  • Must be on the target user’s profile page
  • Target profile must have a “Message” button visible
  • Sending personalized outreach DMs at scale
  • Cold outreach for business, networking, or collaborations
  • Initiating conversations based on profile context
  • Avoiding generic copy-paste DMs
Write a short, friendly and personalized DM for this Instagram user.
Keep it casual, 1-3 sentences max. Do not use hashtags or emojis excessively.
Reference something from their bio if relevant. Be genuine.
Return ONLY the message text.
  • Customize your prompt to match your outreach goal and tone
  • Use blacklist to avoid words that might seem spammy
  • Enable typing simulation for more human-like behavior
  • Keep messages short (1-3 sentences) for higher response rates
  • Start with GPT-4o Mini or DeepSeek Chat for cost efficiency
  • Monitor logs to review what DMs are being sent
  • Use “Go Back After Sending” to chain with other workflow modules