Finch Website (SK)

A novel UX approach to Finch website

Comment

🚀 Finch — Apply Smarter, Not More

Inspiration

Internship recruiting is broken.

Students spend hours jumping between ATS portals, rewriting resumes, and filling out the same forms repeatedly, often with little to no response.

What we noticed is that success isn’t just about effort — it’s about strategy. The students getting interviews are the ones who:

  • Target better-fit roles
  • Tailor their applications
  • Move quickly

We built Finch to bring that strategy to everyone.

Better Outcomes = Targeting + Tailoring + Speed

Finch is designed to help students make every application count, not just apply more.

What it does

Finch is an internship application workflow assistant for engineering, computer science, and quantitative students.

Our current build focuses on a polished frontend experience + AI-powered chatbot, showcasing a unified and intentional application workflow.

With Finch, users can:

  • Discover stronger-fit opportunities (guided UI concept)
  • Generate tailored application materials (simulated flow)
  • Identify ATS keyword gaps (planned feature)
  • Autofill repetitive applications (Chrome extension concept)
  • Track applications in a clean, structured workflow

🤖 AI Chat Assistant

We built a Gemini-powered chatbot directly into the site that:

  • Answers questions about Finch’s features and workflow
  • Guides users through the product
  • Provides contextual, Finch-specific responses (not generic AI output)

How we built it

🧱 Tech Stack

  • Next.js 16
  • React 19
  • TypeScript
  • Tailwind CSS 4
  • Framer Motion (animations)
  • Lucide React (icons)

🎨 Frontend

The frontend is built using reusable, data-driven components, allowing us to iterate quickly on both product and messaging.

We structured the app so that:

  • Features, FAQ, workflow steps, pricing, and team content are all stored in structured data files
  • UI components dynamically render this data across the site
  • The experience feels consistent, scalable, and easy to extend

We focused heavily on:

  • Clean, modern UI
  • Smooth motion and transitions
  • A product-first landing page that feels credible and polished

🤖 Chatbot Architecture

We implemented a chatbot widget connected to a server-side API route.

  • User messages are sent to Gemini 2.5 Flash
  • Each request includes curated Finch-specific context (features, FAQ, product descriptions)
  • This keeps responses grounded and prevents LLM drift

We also built:

  • A fallback response system if the model is unavailable
  • Scoped prompting to ensure concise, relevant answers

Challenges we ran into

Balancing speed vs quality

Most tools optimize for mass applications. We wanted Finch to feel intentional.
Designing for user control instead of full automation was a key challenge.

Defining the scope

Recruiting isn’t one problem — it spans discovery, tailoring, submission, and tracking.
We had to think in terms of a connected workflow, not isolated features.

Communicating trust

We needed the frontend to clearly communicate:

  • Credibility
  • Clarity
  • Differentiation from “spray and pray” tools

What we learned

  • Users want leverage, not replacement
  • Product philosophy matters as much as features
  • Frontend experience strongly shapes perceived quality
  • LLMs are only useful when grounded in real context

What we’re proud of

  • Built from a real problem we’ve lived through
  • Strong, opinionated product direction (quality > quantity)
  • A polished, production-quality frontend
  • An AI assistant that feels useful, scoped, and product-aware

What’s next for Finch

  • Full backend for real job ingestion + tracking
  • Resume parsing + ATS optimization engine
  • Chrome extension for real autofill
  • End-to-end application pipeline

đź’ˇ Final Thought

Students shouldn’t have to choose between quality and efficiency when applying to opportunities that can change their lives.

Finch helps them do both.

Built With

Share this project:

Updates