Claude Climb

A multi-agent career guide designed to empower all students. Through advanced web search and reasoning, it fuses personal priorities, hobbies, and institutional resources into actionable plans.

Comment

Inspiration

The inspiration for Claude Climb came from recognizing the significant challenges college students face when navigating their academic journey and making career choices. We saw that access to clear, personalized guidance and resources is often limited, particularly impacting students from under-resourced communities. We wanted to create a tool that ensures every student's unique strengths, values, and aspirations are heard and used to guide them towards meaningful, sustainable career paths that align with both their skills and societal needs.

What it does

Claude Climb is a multi-agent career guidance system powered by Anthropic’s Claude models. It provides personalized guidance by:

  • Discovering Degree Requirements and Specific Campus Resources: Automatically fetching up-to-date academic pathways, tutoring, advising contacts, and campus support services via Claude web search.

  • Capturing Personal Preferences: Recording student profiles, personality indicators (like MBTI mapped from a 0-100 scale), career priorities, and personal goals/interests.

  • Receiving Tailored Career Recommendations: Analyzing the student's full profile and institutional data to suggest 4-5 suitable careers with match scores and transparent, structured reasoning.

  • Generating Step-by-Step Development Plans: Creating personalized roadmaps—including specific coursework, internships, skills workshops, networking opportunities, and even well-being check-ins—based on the student's university, location, and chosen career path.

How we built it

We built Claude Climb as a multi-agent system using Python and FastAPI. The core components include:

  • Anthropic's Claude Models: Leveraged for the sophisticated reasoning, recommendation, and planning capabilities.

  • Web Search Agent: An agent responsible for fetching and caching relevant external data (degree requirements, resources, internships) specific to the student's institution.

  • Preference Agent: Handles capturing and storing student profile information, including MBTI preferences (on a 0-100 scale), hobbies, natural talents, and life priorities.

  • Reasoning Agent: Analyzes the combined data from the Preference Agent and Web Search Agent to generate career recommendations.

  • Planning Agent: Takes the chosen career and profile data to generate the detailed, actionable roadmap covering academics, extracurriculars, skills, and well-being.

Challenges we ran into

Building a system that effectively fuses diverse data types— from profile data, to web search results, and nuanced preferences like work life balance—presented significant challenges. Ensuring that the numerous Claude Agents worked in harmony to reliably create truly actionable and personalized plans, rather than generic advice, required careful design and parameter tuning. The scheduling and orchestration of the Agents was done meticulously to minimize load times and use of Claude API resources.

Accomplishments that we're proud of

We are proud of the interfaces and functionalities we managed to create through Claude Climb as a two-person team within the span of 20 hours. We are happy with the insights Claude Climb was able to generate, with details down to specific course codes, Professors with relevant research interests, and specific campus labs or resources pulled directly from university data via our Web Search Agent.

What we learned

This project taught us more about the practical application of multi-agent architectures and the capabilities of large language models like Claude for complex reasoning and generation tasks. We learned effective techniques for designing agents with specific responsibilities and managing data flow between them using a shared state. These techniques are indeed more powerful than ones that rely on a single LLM endpoint.

What's next for Claude Climb

The immediate next step for Claude Climb is to significantly enhance the presentation and usability of the career roadmaps generated by our Planning Agent. We plan to implement dynamic flow chart visualizations for these multi-step plans, incorporating roughly estimated timelines for key phases or milestones. This focus on visualization aims to make the detailed, AI-generated plans more accessible, actionable, and encouraging for students navigating their academic and career journey.

Built With

Share this project:

Updates