Low-Code AI Workflow Automation: Integrating With Legacy Systems for Seamless Data Flow

Low-Code AI Workflow Automation: Integrating With Legacy Systems for Seamless Data Flow | Tech Daily Shot

Integrating modern low-code AI workflow automation platforms with legacy systems is one of the most impactful—yet challenging—ways to unlock business value from existing data and processes. As we explored in our complete guide to low-code AI workflow automation platforms, seamless integration is key to building scalable, intelligent workflows that bridge the old and the new.

In this deep-dive, you'll learn how to use a leading low-code AI platform (Make.com) to automate a data flow between a legacy SQL Server database and a modern SaaS CRM (HubSpot)—with AI-powered data enrichment in between. This tutorial offers reproducible, step-by-step instructions, code samples, and troubleshooting tips. If you’re considering other platforms, the principles here apply broadly to most leading low-code solutions.

For more on designing effective prompts and templates, see our sibling article: Prompt Engineering for Low-Code AI Workflow Automation: Templates and Pitfalls.

---

Prerequisites

• Low-Code Platform: Make.com (Free or paid account, tested June 2024)
• Legacy System: SQL Server 2016+ (on-premises or cloud; sample database access)
• Modern SaaS App: HubSpot CRM (Free tier sufficient for contacts API)
• AI Service: OpenAI API key (for enrichment step; GPT-4 or GPT-3.5)
• Connector: ODBC/JDBC or REST API access to legacy system (we’ll use ODBC for SQL Server)
• Basic Knowledge: Familiarity with SQL, API keys, and Make.com workflow basics
• Environment: Windows or Linux machine with sqlcmd and ODBC driver installed

Step 1: Prepare Your Legacy SQL Server for Integration

1. Enable Remote Access (if needed):

   Ensure your SQL Server instance allows remote connections and that TCP/IP is enabled. On the SQL Server host:

2. Create a Read-Only Integration User:

   Open SQL Server Management Studio (SSMS) or use sqlcmd:

   -- In SSMS or via sqlcmd:
   CREATE LOGIN integration_user WITH PASSWORD = 'StrongPassword123!';
   USE YourDatabase;
   CREATE USER integration_user FOR LOGIN integration_user;
   EXEC sp_addrolemember N'db_datareader', N'integration_user';
       

3. Test ODBC Connection:

   Install the ODBC driver and test connectivity:

   
   sudo apt-get install unixodbc odbcinst msodbcsql17
   isql -v "DSN=YourDSN" integration_user StrongPassword123!
       

   If successful, you should see "Connected!" in the terminal.

Step 2: Set Up Your Make.com Scenario

1. Create a New Scenario:

   Log in to Make.com and click Create a new scenario. Give it a descriptive name, e.g., Legacy SQL to HubSpot AI Enrichment.

   Screenshot Description: The Make.com dashboard with the "Create a new scenario" button highlighted.

2. Add the SQL Server Module:

   Search for "SQL Server" in the module list. If not available, use the "HTTP" or "ODBC" module for custom queries.

   Configure the connection using your ODBC DSN, username, and password from Step 1.

   
   SELECT TOP 10 FirstName, LastName, Email, Company FROM dbo.Contacts WHERE Processed = 0;
       

   Set up the module to run this query on a schedule (e.g., every hour).

   Screenshot Description: Make.com scenario canvas with an SQL Server module configured to fetch unprocessed contacts.

Step 3: Add AI Data Enrichment

1. Add OpenAI Module:

   Click the "+" to add a new module after SQL Server. Choose "OpenAI" → "Create a Completion" (or "Chat" for GPT-4).

   Paste your OpenAI API key into the connection dialog.

2. Configure the Prompt:

   Use Make.com’s variable mapping to insert fields from the SQL query. For example:

   
   "Given the following contact info:
   First Name: {{FirstName}}
   Last Name: {{LastName}}
   Company: {{Company}}
   Provide a one-sentence summary of their likely role and business needs."
       

   Set the model to gpt-3.5-turbo or gpt-4.

   Screenshot Description: OpenAI module in Make.com with mapped variables in the prompt.

3. Test the Enrichment:

   Run the scenario once. You should see the OpenAI output appear as a new field in the scenario’s data preview.

Step 4: Push Enriched Data to HubSpot CRM

1. Add HubSpot Module:

   Click "+" and search for "HubSpot." Select "Create or Update a Contact."

   Connect your HubSpot account via OAuth.

2. Map Fields:

   Map FirstName, LastName, Email from the SQL output, and map the OpenAI summary to a custom property (e.g., AI_Enrichment).

   Screenshot Description: HubSpot module field mapping screen in Make.com, showing mapped SQL and AI fields.

3. Optional: Update Legacy System as Processed

   Add another SQL Server (or HTTP/ODBC) module at the end to set Processed = 1 for each contact.

   UPDATE dbo.Contacts SET Processed = 1 WHERE Email = '{{Email}}';
       

Step 5: Schedule, Test, and Monitor Your Workflow

1. Set the Scenario Schedule:

   In Make.com, set your scenario to run on your desired interval (e.g., every hour).

2. Test End-to-End:

   Insert a test record in your SQL Server table:

   INSERT INTO dbo.Contacts (FirstName, LastName, Email, Company, Processed)
   VALUES ('Alice', 'Wong', 'alice.wong@example.com', 'Acme Corp', 0);
       

   Run the scenario. Check HubSpot to confirm the new contact appears with the AI-enriched summary.

3. Monitor Logs & Errors:

   Use Make.com’s scenario log to review any errors or failed runs.

Common Issues & Troubleshooting

• ODBC Connection Fails: Double-check firewall rules, SQL Server authentication mode, and DSN configuration. Test with isql or a database admin tool.
• OpenAI API Errors: Ensure your API key is active, and monitor usage limits. If you see "quota exceeded," request a higher limit or use GPT-3.5 for testing.
• HubSpot Field Mapping: If you don’t see your custom property, refresh the schema in Make.com and confirm the property exists in HubSpot.
• Data Not Appearing in HubSpot: Check for duplicate emails (HubSpot deduplicates by email), and review the scenario run history for mapping errors.
• Legacy Data Format Issues: Use Make.com’s "Text Parser" or "Set Variable" modules to clean or reformat fields before sending them to AI or HubSpot.

Next Steps: Scaling and Securing Your Integration

You’ve now built a robust, AI-powered workflow that connects a legacy SQL Server database to a modern SaaS CRM, with automated enrichment and seamless data flow. This is just the beginning—consider these next steps:

• Enhance Security: Review our Security Best Practices for Low-Code AI Workflow Automation in 2026 to safeguard data in transit and at rest.
• Expand AI Use: Integrate more advanced AI features, such as intent classification or sentiment analysis. See how to build proactive AI customer service workflows for inspiration.
• Automate More Processes: Explore automating other data flows, approvals, or notifications. For a hands-on beginner’s guide, check out How to Build Your First AI-Driven Workflow in a Low-Code Platform.
• Compare Platforms: If you’re evaluating alternatives, see our feature-by-feature comparison of the best low-code AI workflow automation tools.

For a broader strategic perspective and emerging trends, revisit our Pillar: The 2026 Guide to Low-Code AI Workflow Automation Platforms—Build Fast, Scale Smarter.

---

Builder’s Corner: Low-code AI workflow legacy integration is your bridge to future-proofing business processes—no full rewrite needed. With the right tools and a step-by-step approach, you can unlock new value from even your oldest systems.