AI AgentIntermediate10 min read

Build a job-swipe app with Cursor and Grok 4.7

Build a job-card prototype, save roles and draft CV suggestions based on your real experience.

Build a job-swipe app with Cursor and Grok 4.7

What you will build

Job hunting often means switching between open tabs, notes and different versions of your CV. In this tutorial, you will make a small local prototype that shows job cards one at a time, saves the ones you like, and uses Grok 4.7 to suggest CV bullet points based on experience you provide.

The result: a working prototype with sample job listings, a saved list and an AI-assisted draft. You will review and submit applications yourself; this app does not fetch live vacancies or apply on your behalf.

You need: a computer, Cursor or another AI coding assistant, Node.js, an xAI API account, an API key and available API credits. Allow around 30 minutes for the basic interface; setup, troubleshooting, AI integration and review may take longer.

Four stages of the prototype
Four stages of the prototype

Step 1: Decide what the first version should do

Keep the first version small: display a job, save or skip it, and draft CV suggestions for a saved job. Start with three fictional vacancies. Each card should contain a role, company, location, salary if known, and a short description. Leave the salary blank in one example so you can check that the interface says “Not provided”.

Prepare a short fictional or anonymised work history. Avoid using your phone number, address, identity documents, or confidential work information in the first test.

Step 2: Build the job-card interface with Cursor

1. Create an empty folder called job-swipe and open it in Cursor.
2. Open the AI chat and paste this brief. Review any files the assistant proposes before accepting them.

Build a local job browsing prototype using Next.js and TypeScript.
Show one job card at a time with role, company, location, salary and description.
Add accessible Save and Skip buttons; do not require swipe gestures.
Include three fictional sample jobs and show “Not provided” for empty fields.
Let me review saved jobs on the same page and keep them in localStorage.
Provide a README with setup and start commands.
Do not add authentication, live job scraping or an AI API yet.

1. Follow the generated README to install dependencies and start the application. Open the local address printed in the terminal.
2. Save a job, skip another and reload the page. Check that your saved job remains. If setup fails, paste the error message into Cursor and ask it to explain and fix the issue.

Example job card
Example job card

Step 3: Test your CV prompt first

Before adding an API, test the prompt in an AI chat tool you are allowed to use. Replace the example with fictional or anonymised information:

You are a job application writing assistant. Compare the role requirements
with the candidate's real experience and suggest three concise CV bullets.
Use only the facts provided. Never invent employers, qualifications, metrics,
responsibilities or results. Mark missing information as “needs clarification”
and ask up to three follow-up questions.

Role requirements: Help schedule social posts, organise event registrations,
and coordinate with other teams.

My experience: Helped a student society promote an event, organised its signup
list, and coordinated arrangements with three groups.

First list the relevant evidence, then suggest CV bullets.

Check the response against your source material. If the model claims that registrations increased by 30% when you did not supply that figure, remove it. The goal is to present your experience clearly, not invent achievements.

Step 4: Connect Grok 4.7 safely

Use the xAI API with model ID grok-4.7. Check account access, billing, usage limits and current pricing before you start; API use can incur charges.

Create an xAI API key in the xAI console. Keep it in a server-side environment variable; never put it in browser code, screenshots or a public repository. Ask Cursor:

Add a server-side POST /api/draft route to this Next.js prototype.
Read XAI_API_KEY from the server environment and call the official xAI Responses
API with model grok-4.7. Follow the current xAI documentation. Accept only the job description
and the user's supplied experience; limit input length. Ask for three CV bullets
without inventing facts and mark missing details as needing clarification.
The browser must call only our server route and must never receive the token.
Include loading, empty-input and API-error states. Do not submit applications.

Review the generated route and .gitignore. Confirm that the token remains server-side and .env.local is excluded from version control. Try one request with your sample content. If access is denied, check your token permissions, billing and model availability; do not paste your secret into chat.

Server-side model call
Server-side model call

Step 5: Test the whole journey

1. Reload the page and save one of the three fictional jobs.
2. Open it from the saved list, add sample experience and generate suggestions.
3. Check each bullet against the supplied facts and remove unsupported claims.
4. Try an empty input and an unavailable API. Confirm the app shows a helpful error instead of pretending it succeeded.
5. Edit an acceptable draft in your own CV and apply through the employer’s actual application process.

Review checklist
Review checklist

Next steps

Add editing and deletion for saved jobs. Before making the app public, plan privacy protection, authentication, secure server-side storage and abuse prevention. Use vacancy data only when its source permits your intended use. AI suggestions may be wrong, and the prototype does not promise matches or submit applications.

Learn AI, after work

Track your progress, earn XP, and unlock more free tutorials in the AfterWork Academy app.

Open this tutorial in the app

More AI tutorials