Make (Integromat) Integration
Add AI capabilities to Make scenarios using the HTTP module. Build visual automation workflows with intelligent model routing, failover, and cost optimization.
What this enables
- AI-powered steps in any Make scenario via the HTTP module
- Visual workflow builder with drag-and-drop AI processing
- Connect to hundreds of apps and services with AI enrichment
- JSON parsing module handles Antbase responses natively
Quick Start
1. Add an HTTP module
In your Make scenario, add the HTTP > Make a request module.
2. Configure the request
| Field | Value |
|---|---|
| URL | https://antbase.ai/v1/chat/completions |
| Method | POST |
| Body type | Raw > JSON (application/json) |
3. Set headers
Authorization: Bearer your-ant-api-key
Content-Type: application/json4. Set the request body
{
"model": "ant:auto",
"messages": [
{
"role": "user",
"content": "Summarize: {{1.text}}"
}
],
"max_tokens": 1024
}Use Make's variable syntax {{module.field}} to reference data from previous modules.
5. Parse the response
Check Parse response in the HTTP module settings. The AI response text is at choices[1].message.content (Make uses 1-based indexing).
Usage Examples
Lead scoring
{
"model": "ant:fast",
"messages": [
{
"role": "system",
"content": "Score this lead 1-100 based on the data. Return JSON: {"score": N, "reason": "..."}"
},
{
"role": "user",
"content": "Company: {{1.company}}, Title: {{1.title}}, Website: {{1.website}}"
}
],
"temperature": 0,
"response_format": { "type": "json_object" }
}Content generation pipeline
Chain multiple HTTP modules: use ant:fast to generate an outline, then ant:best to write the full article from the outline.
Virtual Models
| Model | Best for | Notes |
|---|---|---|
| ant:auto | General use | Picks optimal model per request |
| ant:fast | Classification, scoring | Low latency for high-volume scenarios |
| ant:best | Complex generation | Frontier models for quality output |
| ant:free | Testing | Zero cost during development |
| ant:reasoning | Analysis | Chain-of-thought reasoning |
| ant:coder | Code tasks | Code-optimized models |
See Virtual Models for the full list.
Troubleshooting
Parse response shows empty
Make sure Parse response is enabled in the HTTP module. The response body is JSON -- set the body type to "Raw" and "JSON" in the module settings.
Timeout on long requests
Make has a default 40-second timeout. For complex prompts, use ant:fast or increase the timeout in the HTTP module's advanced settings.