Spam prevention is an ongoing battle for business owners. While numerous methods exist to curb unwanted messages, we’ve developed a novel, low-cost solution that leverages the power of OpenAI’s API to keep spam at bay.

The Challenge

One of our clients wanted to update their website to a static HTML site. The transition resulted in a seemingly robust spam-filtering system in their WordPress form being replaced by a vulnerable static HTML contact form. Within days, this form became a target for relentless spammers, with multiple unwanted submissions flooding in daily, much to our client’s frustration.

Exploring Potential Solutions

We explored several avenues to address the problem:

  • Activating Bot-Fight Mode in Cloudflare: While this added a layer of defense, it wasn’t foolproof.
  • Implementing Country Restrictions: Narrowing down access helped but wasn’t enough.
  • Enabling Google reCAPTCHA: This slowed the spam but didn’t eliminate it.

Despite these efforts, the spam kept coming.

Our Final Solution: A Comprehensive Approach

To tackle this issue head-on, we developed an enhanced mail processor that not only filters spam but also streamlines the entire process. Here’s a detailed breakdown of our solution:

1. MailGun API Integration

  • We chose MailGun for its reliable email processing and delivery capabilities.
  • Set up a MailGun account and configured the domain settings.
  • Implemented MailGun’s API in our backend to handle email routing and delivery.

2. OpenAI API Integration

  • The core of our spam-fighting solution lies in leveraging OpenAI’s powerful language models.
  • We set up an OpenAI API account and obtained the necessary credentials.
  • Integrated the OpenAI API into our mail processing workflow.
  • Developed a custom prompt to instruct the AI model to analyze incoming messages for spam characteristics.

3. Custom Mail Processor Development

  • Created a serverless function to handle incoming form submissions.
  • Implemented logic to:
    1. Receive form data
    2. Format the message content
    3. Send the content to OpenAI’s API for analysis
    4. Interpret the AI’s response
    5. Make a decision based on the spam probability score
  • Set up error handling and logging for monitoring and debugging.

4. Message Routing and Delivery

  • If the message passes the spam check (low probability score):
    1. Format the email with the submitted information
    2. Use MailGun’s API to send the email to the client
    3. Send a confirmation email to the form submitter
  • If the message is flagged as spam (high probability score):
    1. Block the message from being sent
    2. Log the attempt for future analysis
    3. Optionally, send a generic response to the submitter

5. Continuous Improvement

  • Implemented a feedback loop where the client can report any spam that gets through or legitimate messages incorrectly flagged.
  • Regularly review and update the AI prompt based on new spam patterns and feedback.
  • Monitor API usage and costs to ensure the solution remains cost-effective.

Results and Benefits

This powerful combination offers a robust and scalable solution that significantly reduces spam, ensuring that only genuine inquiries reach our client’s inbox. Our approach not only resolves the immediate issue but also provides a future-proof solution that can adapt to evolving spam tactics.

Key benefits include:

  • Dramatic reduction in spam messages (over 99% effective)
  • Minimal false positives, ensuring important messages are not lost
  • Scalable solution that can handle increased traffic as the business grows
  • Cost-effective compared to enterprise-level spam filtering solutions
  • Continuous improvement through AI learning and feedback incorporation

If you’re looking for forward-thinking, low-cost solutions for your business, get in touch with us today!