Inspiration
Krypt was inspired by a recurring problem we saw across African universities and early-career communities: talented people graduating without access to mentorship, career guidance, or real-time labor market insight. While information exists online, it is fragmented, generic, and often disconnected from local realities.
As builders working at the intersection of AI, education, and workforce development, we saw an opportunity to use generative AI not as a replacement for human mentors, but as a scalable first layer of guidance—one that is accessible, contextual, and affordable for learners who would otherwise be excluded.
What it does
Krypt is an AI-powered mentorship and learning assistant that helps students and early-career professionals:
Navigate career paths and skill requirements
Translate learning into employability outcomes
Receive contextual guidance tailored to their background, goals, and local job market
The system provides structured learning paths, reflective prompts, and actionable feedback bridging the gap between education and employment.
How we built it
Livementor was built using the Gemini API as the core reasoning and generative engine. Gemini’s multimodal and long-context capabilities allow us to:
Analyze user inputs (goals, skills, context)
Generate structured guidance instead of generic answers
Maintain conversational continuity over extended mentorship sessions
We wrapped the model with application logic that enforces:
Guardrails for responsible AI use
Prompt templates aligned with learning science and mentorship research
Human-in-the-loop checkpoints for sensitive or high-impact guidance
The backend handles user profiles, session memory, and feedback loops, while the frontend focuses on simplicity and accessibility for low-bandwidth environments.
Challenges we ran into
One major challenge was avoiding “generic AI advice.” Early versions produced responses that sounded polished but lacked contextual relevance. We addressed this by:
Designing constraint-based prompts
Introducing structured reflection and decision trees
Incorporating local labor-market assumptions instead of global defaults
Another challenge was balancing scalability with responsibility, especially when offering career advice. This pushed us to implement ethical safeguards and clear boundaries on what the system can and cannot recommend.
Accomplishments that we're proud of
Built a working AI mentorship prototype grounded in real user needs
Successfully integrated Gemini API for long-form, contextual reasoning
Designed a system that prioritizes equity, accessibility, and responsible AI
Deployed early pilots with students and young professionals in Nigeria
What we learned
We learned that AI becomes truly powerful in education only when structure, ethics, and context are intentionally designed. Raw model capability is not enough impact comes from how well the system understands people, not just prompts.
What's next for Krypt
Next, we plan to:
Expand localization across African labor markets
Introduce multilingual support
Integrate employer-facing feedback loops
Pilot Krypt within universities and workforce programs at scale
Our long-term goal is to make high-quality mentorship a public utility, not a privilege.
Log in or sign up for Devpost to join the conversation.