Automating Accessibility Outreach: A Powerful Workflow with n8n, Perplexity, and Gemini

Website accessibility is crucial for creating an inclusive online environment. However, reaching out to businesses with accessibility issues can be a time-consuming and manual process. We decided to tackle this challenge head-on by building an automated workflow that leverages the power of n8n, Perplexity AI, Gemini AI, and our own website scanner. This is how we did it.

The Challenge: From Promotional Emails to Actionable Insights

Our journey began with a mountain of promotional emails cluttering our employees' inboxes. We knew that hidden within these emails were hundreds of potential webshops that could benefit from our accessibility testing services. For this project, we wanted to try a different approach rather than using a standard subscription-based CRM or email service like Brevo. The challenge was to efficiently identify these webshops, assess their accessibility, and then reach out to them with personalized and actionable feedback using our own custom-built workflow.

Our Modular n8n Architecture

Instead of one monolithic workflow, we created a series of distinct, interconnected n8n projects, each with a specific job. This modular approach made the system more manageable, scalable, and easier to debug.

1. Initial Data Funnel: Identifying Webshops with AI

The first step was to sift through promotional emails. We used an n8n workflow connected to Gemini AI to analyze email content and identify potential webshops. This process gave us our initial list of about 150 websites, which were then fed into our Supabase database.

2. Core Workflows: The Heavy Lifting

Once a website was in our system, a series of automated workflows took over:

• Accessibility Scoring: A dedicated n8n project would take a website URL from the database, run it through our custom website accessibility scanner, and update the database entry with the resulting score.
• Contact Discovery: Another workflow was triggered, which used Perplexity AI to find the best contact email address for the webstore. This email was then saved alongside its accessibility score in Supabase.
• Personalized Email Generation: A third workflow would activate once a score and email were present. It used ChatGPT and CursorAI to select and customize the appropriate email template based on the store's type and accessibility score.

3. Automation Triggers and Team Integration

The real power of our system came from how we triggered these workflows and integrated them with our team's daily tools:

• Scheduled Database Checks: A recurring n8n task constantly monitored our Supabase table for new website entries, automatically kicking off the scoring and contact discovery process.
• Flexible Team Integration: We created a simple but powerful integration point. While we used Slack & Discord, the system is designed so that other tools can easily be connected. This allows team members to add new websites directly from their preferred communication channel, and it enables us to send notifications or get data from users through various platforms without them needing direct access to n8n.

4. The Final Step: Controlled Outreach

The email generation workflow was connected to n8n's Gmail connector, giving us valuable flexibility. Depending on the campaign, we configured the workflow to either send the personalized emails directly or to create drafts in Gmail. This second option allowed for a final human check on the generated emails before they were sent out, ensuring quality and accuracy.

The Results and Key Takeaways

By breaking down the process into multiple, specialized n8n projects, we created a highly effective and scalable system for accessibility outreach. This automated process saved us countless hours of manual work and allowed us to reach out to a large number of webshops with personalized and actionable feedback.

This project demonstrates the incredible potential of using a combination of AI and automation tools to solve real-world problems. By thinking creatively and leveraging the right technologies, we were able to turn a cluttered inbox into a powerful engine for positive change. We plan to continue with this approach, integrating and automating different flows that are currently done manually, but can be improved with the help of tools like n8n, Gemini, and ChatGPT.

What's Next? More Ideas for Your Automation Journey

The workflow we built is just one example of what's possible. This same combination of automation and AI can be adapted to solve countless business challenges. Here are a few other complex, high-value flows we can help customers set up:

• Automated Lead Enrichment & Sales Outreach: When a new lead signs up, a workflow can automatically enrich their profile with data from the web (like their company size, role, and recent news). An AI can then generate a personalized "ice-breaker" sentence, update your CRM, and create a task for your sales team, delivering all the key info directly to them in Slack.
• Intelligent Customer Support Triage: Automatically analyze incoming support tickets using AI to determine the topic, urgency, and customer sentiment. The workflow can then tag the ticket, route it to the correct department (e.g., Billing, Tech Support), and send an initial reply with links to relevant help articles, all before a human agent even sees it.
• E-commerce Competitor Price Monitoring: Set up a daily workflow that automatically scans your competitors' websites for the prices of key products. This data can be logged in a database, and the system can alert you if a competitor's price drops below a certain threshold, allowing you to react quickly and stay competitive.

If you have a manual, repetitive process in your business, there's a good chance it can be automated. We'd love to help you explore the possibilities.