n8n

n8n is an open-source workflow automation platform. Because AICredits is OpenAI-compatible, you can use it inside any n8n node that accepts a custom base URL — including the OpenAI node, the AI Agent node, and the HTTP Request node. This gives your automations access to Claude, Gemini, DeepSeek, and 300+ models with a single INR-billed API key.

Overview

There are three ways to use AICredits in n8n:

OpenAI Node — Custom Credentials

The n8n OpenAI node supports a custom base URL via OpenAI API credentials. Set it once and every OpenAI node in your workspace automatically uses AICredits.

1. Open Credentials → New → OpenAI API — Go to Settings → Credentials → Add Credential → OpenAI API.
2. Set the API Key — Paste your AICredits API key (starts with sk-) into the API Key field.
3. Set a custom Base URL — Expand Additional Fields and set Base URL to https://api.aicredits.in/v1.
4. Save and use in any OpenAI node — Select these credentials in any OpenAI node. In the Model field, type the AICredits model ID directly, e.g. anthropic/claude-sonnet-4.5.

HTTP Request Node

Use the HTTP Request node for full control — useful for image generation, checking your credit balance, or passing custom parameters the OpenAI node does not expose.

Chat Completions:

Method: POST
URL: https://api.aicredits.in/v1/chat/completions

Headers:
  Authorization: Bearer {{ $credentials.aicreditsApiKey }}
  Content-Type: application/json

Body (JSON):
{
  "model": "openai/gpt-4o-mini",
  "messages": [
    { "role": "system", "content": "You are a helpful assistant." },
    { "role": "user", "content": "{{ $json.user_message }}" }
  ],
  "temperature": 0.7,
  "max_tokens": 512
}

Check Credit Balance:

Method: GET
URL: https://api.aicredits.in/v1/credits

Headers:
  Authorization: Bearer {{ $credentials.aicreditsApiKey }}

# Response:
{
  "balance_inr": 245.50,
  "currency": "INR"
}

Image Generation:

Method: POST
URL: https://api.aicredits.in/v1/images/generations

Headers:
  Authorization: Bearer {{ $credentials.aicreditsApiKey }}
  Content-Type: application/json

Body (JSON):
{
  "model": "dall-e-3",
  "prompt": "{{ $json.image_prompt }}",
  "size": "1024x1024",
  "quality": "standard",
  "n": 1
}

# Access the URL in subsequent nodes via:
# {{ $json.data[0].url }}

AI Agent Node

n8n's AI Agent node (under the LangChain section) supports a custom OpenAI-compatible chat model. This lets you build autonomous agents inside n8n that call tools like web search, database queries, or other n8n nodes.

1. Add an AI Agent node — In your workflow, add the AI Agent node from the LangChain category.
2. Add a Chat Model sub-node — Click + on the Chat Model input → select OpenAI Chat Model → choose your AICredits credential and set the model ID.
3. Add tools — Connect tool nodes (Wikipedia, Calculator, HTTP Request, etc.) to the Tools input of the agent.
4. Set the system prompt and input — Configure the agent's system prompt and pass the user's input message from a previous node.

Example Workflows

Classify & Route Messages

Classify incoming Telegram/WhatsApp messages and route them to different downstream nodes:

# Workflow: Telegram Trigger → OpenAI Node → Switch Node

# OpenAI Node config:
Operation: Message a Model
Model: openai/gpt-4o-mini
System Prompt: |
  Classify the user message into exactly one of these categories:
  billing, technical, sales, general
  Reply with only the category name — no other text.
User Message: {{ $json.message.text }}

# Switch Node routes on {{ $json.message.content }}:
- billing   → Billing Support workflow
- technical → Tech Support workflow
- sales     → Sales CRM workflow
- general   → FAQ auto-reply

Summarise Emails

Trigger on new Gmail messages and generate a one-sentence summary to post in Slack:

# Workflow: Gmail Trigger → OpenAI Node → Slack Node

# OpenAI Node:
Operation: Message a Model
Model: anthropic/claude-haiku-4.5-20251001
System Prompt: Summarise the email in one sentence. Start with the sender's name.
User Message: |
  From: {{ $json.from.value[0].address }}
  Subject: {{ $json.subject }}

  {{ $json.text }}

# Slack Node:
Channel: #email-digest
Message: 📧 {{ $node["OpenAI"].json.message.content }}

Extract Structured Data

Extract structured fields from unstructured text (invoices, forms, tickets) and write to Google Sheets:

# Workflow: Webhook → OpenAI Node → Google Sheets Node

# OpenAI Node:
Operation: Message a Model
Model: openai/gpt-4o-mini
System Prompt: |
  Extract the following fields from the text and respond with JSON only:
  {
    "vendor_name": string,
    "invoice_number": string,
    "amount": number,
    "currency": string,
    "due_date": "YYYY-MM-DD"
  }
  If a field is missing, use null.
User Message: {{ $json.invoice_text }}

# Parse the JSON in a Code node:
# const data = JSON.parse($input.first().json.message.content);
# return [{ json: data }];

Tips

• Use expression mode for dynamic model selection — Set the Model field to an expression like {{ $json.preferred_model || 'openai/gpt-4o-mini' }} to pick models dynamically per workflow run.
• Add a low-balance check before expensive nodes — Use the HTTP Request node to GET /v1/credits and add an IF node — stop the workflow if balance_inr < 10 to avoid failed requests.
• Use JSON mode for extraction workflows — Enable Response Format: JSON to guarantee parseable output. Combine with a Code node to safely parse and pass fields downstream.
• Store the API key as an n8n credential — Never paste the API key directly into node parameters. Create an OpenAI API credential once and reference it via {{ $credentials.aicreditsApiKey }} for secure, centralised key management.
• Switch models mid-workflow for cost optimisation — Use gpt-4o-mini for classification/extraction (cheap, fast) and claude-sonnet only for final response generation (quality). This reduces cost by 80–90% vs using a frontier model throughout.