NextStep

Inspiration:

Career orientation is often a source of anxiety and inefficiency for students. Our starting point was a simple observation: the problem lies in a lack of information, a lack of methodology, and unequal access to networks.

Many students have ambition but no concrete idea of the path to get there. Furthermore, finding the right contacts or internships often depends on one's social background. We wanted to create a tool that "hacks" this social determinism by transforming the career orientation process into an exact science, accessible to all.

How does it work?

NextStep is an intelligent web platform that transforms a vague ambition into a personalized action plan.

Initial Profiling: Upon their first login, the student enters their academic history (school, major), past experiences (internships, volunteering), and their study location. This data allows the algorithm to contextually tailor future recommendations geographically and academically.

Mode Selection: The user chooses between two paths based on the maturity of their professional project:

"Defined Project" Mode: The user enters their target job. The app displays a detailed job description, suggests related professions to broaden horizons, and generates a strategic roadmap (target schools, required internships, etc.) to maximize their chances. This roadmap is interactive: each major step (e.g., "Find an internship in Finance") is clickable and breaks down into concrete, localized sub-tasks (e.g., "Register for the Paris Finance Forum").

"Undecided Project" Mode: An AI Chatbot engages the user to identify their values and interests. It then proposes targeted Sector Files. These files act as bridges, containing links to specific roles that, once selected, trigger the creation of a personalized roadmap.

Roadmap and Connections: An AI analyzes the profile data and the user's LinkedIn profile to:

Activate the Network: Identify relevant contacts (alumni, distant family members identified via last names) working in target companies. These contacts are accessible directly in the "Network" tab.

Update the Roadmap: Suggest internships made accessible through the identified network and propose relevant events (forums, conferences) based on the student's geography and current date.

Dynamic Chatbot Assistant: A dedicated tab allows users to chat with the assistant to update their roadmap and Network tab in real-time as their journey evolves.

How we built it:

To build this project rapidly, we centralized everything around three core tools:

Lovable: The backbone of our project. We didn't code the interface line-by-line; we used Lovable to generate the entire application (Frontend and logic) using natural language prompts. This allowed us to achieve a clean, functional site in record time.

Featherless AI: To power the system's intelligence, we integrated the Featherless API. It drives our orientation Chatbot and, crucially, dynamically generates the career roadmaps by transforming user requests into structured action plans.

Firecrawl: For the "Data" layer, we used Firecrawl. This powerful scraping tool allows us to scan the web and transform complex web pages into clean, usable data for our AI, specifically for networking and internship research.

Challenges we ran into:

Managing Career Pivots: We initially struggled to make the roadmap flexible. Early on, if a user changed their mind (e.g., moving from "Finance" to "Gardener"), the AI kept remnants of the previous job, creating incoherent roadmaps. We had to refine the AI's memory management to distinguish between a radical reset and a soft transition where "soft skills" could be preserved.

Network Profile Relevance: This was the toughest part. Initially, the tool suggested irrelevant people (homonyms) or untraceable alumni. The AI tended to "hallucinate" family ties or alumni connections. We spent significant time refining search criteria to ensure suggested contacts were real, relevant individuals.

Accomplishments that we're proud of:

The "Relational Matching" Engine: we successfully transformed passive data (LinkedIn, last names, schools) into active social capital.

Full Account System: Beyond a prototype, we implemented a functional authentication system. Users have a secure personal space that saves their progress, conversation history, and unlocked contacts in real-time.

Real-time Adaptability: The entire environment—Roadmap, events, and Network—reconfigures instantly if a user decides to "pivot" their professional goal.

What we learned:

The Art of the MVP: The intense hackathon format forced us to prioritize "core" features that deliver immediate value.

Frontend Acceleration: Discovering Lovable radically changed our approach to UI, allowing us to iterate at lightning speed.

AI Integration via Featherless: We learned how to deploy powerful language models without heavy infrastructure, making our chatbot responsive and relevant.

What's next for NextStep:

School/Corporate Partnerships: Connecting the platform directly to school intranets to access verified alumni directories.

Mentoring Module: Allowing users to contact mentors directly through the platform.

"One-Click Apply": The ability to apply for found internships directly from the NextStep interface.