LinkedIn Agent

Installation

Install peer dependencies:

npm install openai zod

Copy the code of Linkedin Tool and paste it in `utils/linkedinTool.ts`

Copy the code of ChatOpenAI and paste it in `utils/chatOpenAi.ts`

Add Environment Variables

.env

OPENAI_API_KEY = "YOUR_SAMPLE_API_KEY";
PROXYCURL_API_KEY = "YOUR_SAMPLE_API_KEY";
/* You can get one from - https://platform.openai.com/api-keys */
/* You can get one from - https://nubela.co/proxycurl/linkedin */

Now copy the backend code of Agent

Add the following code to your /api/linkedinAgent/route.ts file:

route.ts

import { NextRequest, NextResponse } from "next/server";
import { LinkedinTool } from "@/utils/linkedinTool";
import { ChatOpenAI } from "@/utils/chatOpenAi";
import { z } from "zod";
 
const requestBodyValidator = z.object({
  url: z.string(),
  type: z.enum(["profile", "company"]),
});
 
const validateRequestBody = (body: any) => {
  return requestBodyValidator.parse(body);
};
 
const handleProfile = async (
  url: string,
  chatOpenAI: ChatOpenAI,
  tool: LinkedinTool
) => {
  const profileData = await tool.searchProfile(url);
  const content = JSON.stringify(profileData);
  const data = await chatOpenAI.chat({
    prompt: `Giving you a data of a linkedin user profile. Analyse and give me a detailed summary of the profile in a HTML tabular format.`,
    context: content,
    outputFormat: profileOutputResponse,
  });
  return data;
};
 
const handleCompany = async (
  url: string,
  chatOpenAI: ChatOpenAI,
  tool: LinkedinTool
) => {
  const companyData = await tool.searchCompany(url);
  const content = JSON.stringify(companyData);
  const data = await chatOpenAI.chat({
    prompt: `Giving you a data of a linkedin company profile. Analyse and give me a detailed summary of the company in a HTML tabular format.`,
    context: content,
    outputFormat: companyOutputResponse,
  });
  return data;
};
 
export async function POST(request: NextRequest) {
  try {
    const body = await request.json();
    const { url, type } = validateRequestBody(body);
 
    const tool = new LinkedinTool({
      apiKey: process.env.PROXYCURL_API_KEY,
    });
 
    const chatOpenAI = new ChatOpenAI({
      apiKey: process.env.OPENAI_API_KEY,
      model: "gpt-4o",
    });
 
    let data;
    if (type === "profile") {
      data = await handleProfile(url, chatOpenAI, tool);
    } else {
      data = await handleCompany(url, chatOpenAI, tool);
    }
 
    return NextResponse.json({ success: true, data: data });
  } catch (err: any) {
    console.error("Error in POST: ", err);
    return NextResponse.json({
      success: false,
      status: 500,
      error: err.message,
    });
  }
}
 
const companyOutputResponse = `<table>
  <thead>
    <tr>
      <th>Field Name</th>
      <th>Description</th>
    </tr>
  </thead>
  <tbody>
    <tr>
      <td>Name</td>
      <td>"name of company + tagline if any"</td>
    </tr>
    <tr>
      <td>Description</td>
      <td>"description of company if any"</td>
    </tr>
    <tr>
      <td>Locations</td>
      <td>
        * Loc 1<br>
        * Loc 2<br>
        ... if any<br>
        * HQ country + state + city<br>
        * founded year + date
      </td>
    </tr>
    <tr>
      <td>Head</td>
      <td>
        * company size<br>
        * company followers
      </td>
    </tr>
    <tr>
      <td>Stats</td>
      <td>
        * company 1<br>
        * company 2<br>
        ... if any
      </td>
    </tr>
    <tr>
      <td>Similar Companies</td>
      <td>
        * Acc 1<br>
        * Acc 2<br>
        * Acc 3... if any
      </td>
    </tr>
    <tr>
      <td>Accomplishments</td>
      <td>
        * key 1<br>
        * key 2<br>
        * key 3... if any
      </td>
    </tr>
    <tr>
      <td>Key Takeaways</td>
      <td>
        * Suggestion 1<br>
        * Suggestion 2<br>
        * Suggestion 3... if any
      </td>
    </tr>
    <tr>
      <td>Suggestions</td>
      <td>
        * Note 1<br>
        * Note 2<br>
        * Note 3... if any
      </td>
    </tr>
  </tbody>
</table>`;
 
const profileOutputResponse = `<table>
  <thead>
    <tr>
      <th>Field Name</th>
      <th>Description</th>
    </tr>
  </thead>
  <tbody>
    <tr>
      <td>Name</td>
      <td>"name of user if any"</td>
    </tr>
    <tr>
      <td>Occupation</td>
      <td>"occupation if any"</td>
    </tr>
    <tr>
      <td>Location</td>
      <td>"city + state + full country name if any"</td>
    </tr>
    <tr>
      <td>Summary</td>
      <td>"summary of headline + occupation + summary if any"</td>
    </tr>
    <tr>
      <td>Experiences</td>
      <td>* Exp 1<br>* Exp 2<br>... if any</td>
    </tr>
    <tr>
      <td>Education</td>
      <td>* Edu 1<br>* Edu 2<br>... if any</td>
    </tr>
    <tr>
      <td>Statistics</td>
      <td>* folower count<br>* connection count<br>... if any</td>
    </tr>
    <tr>
      <td>Accomplishments</td>
      <td>* Acc 1<br>* Acc 2<br>... if any</td>
    </tr>
    <tr>
      <td>Key Takeaways</td>
      <td>* Key Takeaway 1<br>* Key Takeaway 2<br>... if any</td>
    </tr>
    <tr>
      <td>Suggestions</td>
      <td>* Suggestion 1<br>* Suggestion 2<br>... if any</td>
    </tr>
  </tbody>
</table>`;

Usage

Initialize FrontEnd

Create a frontend and perform a fetch call to get the data from backend.

 const settings = {
      method: "POST",
      headers: {
        Accept: "application/json",
      },
      body: JSON.stringify({ url: profileUrl, type }),
    };
    const fetchResponse = await fetch(`/api/linkedinAgent`, settings);
    const data = await fetchResponse.json();
    const responseData = data.data.output
      .replace("```html\n", "")
      .replace("\n```", "");
    console.log(responseData);

Now you can customize it to suit your needs and continue building on top of it. In this case, we are receiving the output in HTML table format. You can modify this by changing the outputFormat prop of ChatOpenAI.