Make.com + Ertas

Connect Make.com automation scenarios to Ertas-trained local models instead of expensive cloud AI APIs, eliminating per-token costs while keeping your existing workflow logic intact.

Overview

Make.com (formerly Integromat) has established itself as the leading no-code automation platform, powering hundreds of thousands of scenarios across businesses of every size. AI agencies in particular rely on Make.com to orchestrate complex workflows — pulling data from CRMs, processing documents, generating content, and routing outputs to downstream tools. Its visual scenario builder makes it accessible to non-developers while remaining powerful enough for sophisticated multi-step automations that would otherwise require custom code.

The challenge for agencies running AI-powered scenarios at scale is cost. Every AI step in a Make.com scenario typically calls a cloud API like OpenAI or Anthropic, incurring per-token charges that compound quickly across hundreds of daily executions. A single content-generation scenario processing 500 requests per day can easily rack up thousands of dollars in monthly API fees. Worse, rate limits from cloud providers can throttle high-volume scenarios during peak hours, causing execution delays and failed runs that require manual intervention.

How Ertas Integrates

Ertas bridges the gap between no-code automation and cost-effective AI by letting you train purpose-built models that run on your own infrastructure. Using Ertas Studio, you fine-tune a model on your specific use case — whether that is customer email classification, product description generation, or lead scoring — and export it as a GGUF file. Deploy the model through Ollama on any server or local machine, and you immediately have an OpenAI-compatible API endpoint ready to receive requests from Make.com.

The integration requires no changes to your existing scenario logic. Make.com's HTTP module can call any REST endpoint, so you simply swap the cloud API URL for your local Ollama address. The request and response formats match the OpenAI chat completions spec, meaning your existing JSON templates, response parsing, and error handling all continue to work identically. The only difference is that inference now runs on hardware you control, at a fixed monthly cost regardless of how many tokens you process.

Getting Started

1
Fine-tune your model in Ertas Studio
Upload your training data — previous AI outputs, curated examples, or domain-specific documents — and configure a fine-tuning job in Ertas Studio. Select a base model appropriate for your use case and set your training parameters.
2
Export model in GGUF format
Once training completes, download your fine-tuned model as a GGUF file with the quantization level that balances quality and speed for your hardware. Ertas provides the accompanying Modelfile for Ollama deployment.
3
Deploy with Ollama
Register the model with Ollama using the provided Modelfile and start the inference server. The OpenAI-compatible API will be available on port 11434, ready to accept chat completion requests.
4
Configure the Make.com HTTP module
In your Make.com scenario, add or update an HTTP module to point to your Ollama endpoint. Set the URL to your server address, configure the JSON request body with the model name and messages array, and map your scenario variables into the prompt.
5
Test and activate the scenario
Run the scenario in test mode to verify responses match your expectations. Compare output quality against the previous cloud API results, then activate the scenario for production use. Monitor execution logs for any formatting differences.

json

{
  "url": "http://your-server:11434/v1/chat/completions",
  "method": "POST",
  "headers": {
    "Content-Type": "application/json"
  },
  "body": {
    "model": "my-ertas-model",
    "messages": [
      {
        "role": "system",
        "content": "You are a customer support classifier."
      },
      {
        "role": "user",
        "content": "{{1.email_body}}"
      }
    ],
    "temperature": 0.3,
    "max_tokens": 256
  }
}

Make.com HTTP module configuration pointing to a local Ollama endpoint serving an Ertas-trained model. The {{1.email_body}} placeholder maps to a previous module's output.

Benefits

Eliminate per-token API costs — pay only for your server hardware regardless of volume
No rate limits or throttling during high-volume scenario executions
Full data sovereignty — customer data processed in Make.com scenarios never leaves your infrastructure
Keep your existing scenario logic, JSON templates, and response parsing unchanged
Fixed monthly infrastructure cost makes budgeting predictable for agency clients
Fine-tuned models produce more consistent, domain-specific outputs than generic cloud APIs