Skip to main content
As we already know from About MCP — MCP allows to LLM use API by just asking AI. This approach is especially useful for connecting tools like Generect, which provides live company and lead data, to OpenAI’s Agent Builder, which orchestrates multi-step workflows.

​
What Generect does

Generect Live API delivers real-time business and contact data from verified sources like LinkedIn, Crunchbase, and company websites. Unlike static databases, it doesn’t rely on outdated snapshots – every request fetches live information. What makes it useful:
  • Fresh data: Always current, no batch updates needed.
  • AI email finder: Predicts and validates emails with a proprietary system.
  • Catch-all validation: Detects risky addresses with far higher accuracy.
API Base URL: https://api.generect.com/

​
Why combine Generect + OpenAI Agent Builder

OpenAI’s Agent Builder lets you visually design, debug, and deploy multi-step AI workflows.
By integrating Generect into it via MCP, you can create agents that search live company data, find leads, validate emails, and summarize results, all in one conversational flow. Imagine typing this into a chat: “Find 5 leads from Google and 5 from Microsoft, then compare their recent hires.” The agent automatically:
  1. Calls Generect through MCP to fetch live data.
  2. Compares results.
  3. Returns a clear summary.

​
How the integration works

Connecting Generect to OpenAI’s Agent Builder through MCP only takes a few minutes. Once it’s connected, your agents can directly call Generect’s live API (fetching company data, finding leads, and validating emails) all from within your workflow. Follow these steps:

​
Step 1. Add an MCP Node

In your workflow canvas inside Agent Builder, click “Add Node” and select MCP from the list. Agent builder step 1 This node acts as the bridge between your agent and the Generect API.

​
Step 2. Click ”+ Server”

In the MCP node settings panel, click the ”+ Server” button to add a new server connection. Agent builder step 2 This tells Agent Builder which external service you want to connect to.

​
Step 3. Enter the Generect API URL

In the Server URL field, enter: https://api.generect.com/ Agent builder step 3
Make sure the URL ends with a trailing slash / because the connection won’t work correctly without it.

​
Step 4. Choose authentication type

From the authentication dropdown, choose Access Token. Generect uses a simple Bearer Token (API key) system for secure access. Now you’ll need to generate that token.

​
Step 4.1. Log In to Generect

Go to https://beta.generect.com/ and sign in with your account.

​
Step 4.2. Open the API section

In the top navigation bar, find the API option and click it. Agent builder step 4.2

​
Step 4.3 Create a new key

Click ”+ New Key” to generate a new access token. Agent builder step 4.3

​
Step 4.4 Paste the key

After creation, copy your new key and return to Agent Builder. In the MCP server connection window:
  • Paste the key into the Access Token field.
  • Review your settings.
  • Click Connect.
Once connected, you’ll see your Generect MCP server appear in the list. Agent builder step 4.4

​
Step 5. Create your agent workflow

In Agent Builder, everything happens on a visual canvas. You’ll build a workflow with a series of nodes that represent steps. A simple workflow might look like this: User Input → Agent → MCP (Generect) → Transform → Output Here’s what each node does:
  • Start Node: Captures user input (e.g., “Find founders from biotech startups.”)
  • Agent Node: Interprets the query, decides what to fetch.
  • MCP Node: Sends a request to the Generect API.
  • Transform Node: Reshapes or filters the data.
  • Output Node: Returns clean, readable results to the user.

​
Step 6. Configure the MCP Node

When adding an MCP node in Agent Builder:
  1. Choose your server: generect
  2. Select the endpoint you want to use, for example:
    • /companies/by_icp/
    • /leads/by_icp/
    • /emails/email_validation/
Add parameters. You can use variables from previous nodes, like: 
{
 "industry": {{ user_input.industry }},
 "location": {{ user_input.location }},
 "limit": 5
}
  1. Test your call using Preview to confirm the data looks correct.

​
Step 7. Transform and display results

Once the MCP node fetches data, use a Transform node to shape it into an easy-to-read format. For example, you might only want to show names, companies, and validated emails. Sample transformation (pseudo-code): 
{
  "leads": [
    {  
      "name": lead.full_name,
      "company": lead.company_name,
      "email": lead.valid_email
    }
  ]
}
Then pass that to your final Agent node to generate a summary like: “Here are 5 verified leads from biotech startups in San Francisco.”