PlexiPass

PlexiPass helps you understand, prepare, and navigate complex visa processes – powered by Perplexity.

Comment

Inspiration

I had many acquaintances who were in the process of applying for visas for their higher education. However, I found the process to be really complex, and doing it all by oneself without an agent’s help seemed almost impossible. I wanted to make the process easier—for my friends and for everyone seeking to get a visa for any country, for any purpose. PlexiPass is made to be as customisable as possible to cover all the dynamics of visa applications.

What it does

PlexiPass is a smart visa assistant powered by Perplexity APIs. Users can chat with the assistant for generic questions related to visas. More importantly, they are provided with custom features like generating a full report on their Visa Success Probability, which takes into account various details like age, work experience, ties to home, and more. Multiple such custom features are available, which fetch the latest rules and regulations applicable to the user's case and generate a personalized dashboard report that can be downloaded as an HTML file—or else, it stays in the user's account data.

Key Features:

  • Generic visa-related questions answered instantly
  • Visa pathway generator based on user profile
  • Visa interview preparation guide tailored to the user
  • Visa success probability predictor
  • Personal visa coach for step-by-step planning
  • Chat quota renews daily, while custom features (deep research) quota renews monthly for each user

How we built it

  • Next.js for full-stack web app, routing, SSR, and server APIs
  • Perplexity Sonar API for AI chat, document search, and visa Q&A
  • Tailwind CSS for modern, responsive UI components
  • Supabase as our primary PostgreSQL-based user DB
  • Clerk.dev for user authentication (Email, Google)
  • Vercel for hosting, CI/CD, and performance optimisation (Speed Insights, Analytics)
  • Paddle for payments to support the freemium subscription model

For the Perplexity APIs, we utilised three different models for our use cases. For having a generic visa-related chat, we used the "sonar-pro" model. We wanted to generate a user-friendly title for each chat, so we used the "sonar" model for that. For custom features like generating visa pathways and more, we used the "sonar-deep-research" model and prompted it to return results in our desired JSON schema to be rendered for the user.

Challenges we ran into

There were a few challenges I faced while developing PlexiPass, including:

  • CORS issues when integrating external APIs securely, specifically with Clerk Auth
  • Figuring out the json_schema parameter in the Perplexity API call and syncing it with our application
  • Handling all edge cases and ensuring every component works smoothly

Accomplishments that we're proud of

I'm very proud to have launched PlexiPass as a SaaS-level application with a fully integrated end-to-end user flow. I'm more than happy to be able to help my friends—and many others—in their visa application journey :) Visas need not be as complex as they currently are, especially for students. The integration of Perplexity AI boosts the capabilities of PlexiPass.

What we learned

I learned a lot throughout the development of PlexiPass. Some major takeaways include:

  • Real-world prompt engineering for structured outputs (like JSON) is as hard as writing code
  • Syncing authentication between frontend and backend (Clerk with SSR) requires precision
  • AI performs much better when you restrict its domain with a focused system prompt
  • Seamless UI/UX integration is essential for a properly working subscription model and user data management

What's next for PlexiPass

I have many future plans for PlexiPass, as I want to launch it in the market as a SaaS made for visa applicants to make their process easier. I also want to add features that could appeal to visa agents, who can benefit from PlexiPass as well. My major upcoming plans include:

  • Fine-tuning existing prompts and strengthening the security layer to prevent unauthorized access
  • Integrating a production-ready payment model
  • Introducing an AI-powered document validator and checker
  • Implementing mobile-first design and PWA (Progressive Web App) support

Built With

Share this project:

Updates