Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.tuco.ai/llms.txt

Use this file to discover all available pages before exploring further.

​
AI Message Generation

Instead of writing a fixed template, give the AI a prompt describing what to say. The AI generates a unique, personalized message for each lead.

​
How It Works

  1. In the campaign builder, toggle a step to AI mode
  2. Write a prompt: 
    Write a friendly follow-up to {firstName} at {companyName}.
Mention that we help companies like theirs reduce outreach costs
by 60% with iMessage automation. Keep it under 3 sentences.
  3. The AI uses the lead’s data (name, company, tags, custom fields) to generate a personalized message
  4. Each lead gets a different message — not a template with variables swapped in

​
Fallback Safety

Every AI step has a manual fallback message. If AI generation fails (rate limit, timeout, model error), the fallback is sent instead. You always have a safety net.

​
Cost & Billing

AI message generation is tracked per workspace:
  • Token usage logged to ai_usage collection
  • Visible in Activity Logs → AI Usage tab
  • Uses GPT-4o-mini by default for cost efficiency

​
AI Tools in Campaigns

​
Workspace API Tools

Create reusable API integrations that AI can call during message generation:
  1. Go to Integrations → API Tools
  2. Create a tool (e.g. “Check Credit Score”)
  3. Configure: URL, method, headers, body, response mapping
  4. Use in campaigns as variables: {api_check_credit_score}

​
AI + Branching

Combine AI-generated messages with data-driven branching:
  • Step 1: AI sends personalized outreach
  • Branch: Did the lead reply? → Yes path / No path
  • Step 2 (no reply): AI sends a follow-up referencing the original message
  • Branch: Check HubSpot property → Route to different paths based on deal stage

​
Testing Campaigns

The Test Conversation panel lets you simulate the full campaign for a real lead:
  1. Select a lead — uses their real data (name, phone, tags, CRM properties)
  2. See each step — AI-generated messages use real AI calls (billed to your usage)
  3. Reply as the lead — type what the lead would say
  4. See stop conditions — green/red pills show which conditions passed or triggered
  5. See branches — “Evaluate live” checks real HubSpot properties, GHL tags, and custom API calls
  6. Fast-forward waits — skip “3 days later” to see what happens next
  7. Path simulation — visual timeline showing every step, wait, branch, and the delivery method (iMessage vs SMS)
  8. CRM links — open the lead in HubSpot or GHL to change data mid-test, then restart
AI-generated test messages count toward your AI usage. Use sparingly on expensive models.