student • founder • developer

building markwise - from scratch to private beta

back in march 2025, i started with just an idea: could ai genuinely improve revision for students? just a few months later, markwise was already helping students in a private beta.


where it all started

revision has always been a pain. traditional methods felt inefficient and outdated. my goal was simple: build something students actually wanted to use.

here were my initial ideas:

  • personalized practice: revision that adapts in real-time based on your performance.
  • exam-specific insights: leveraging examiner reports and past papers to deliver targeted practice.
  • instant, actionable feedback: no more waiting - get immediate insights to correct mistakes quickly.

from concept to reality

the early days

i kicked off with a simple prototype, focusing on the core ai functionality:

  • built an initial model to match exam questions to student weaknesses.
  • created a straightforward ui using next.js and tailwind, prioritizing simplicity and ease of use.

developing the ai tools

building effective ai for education was challenging but rewarding:

  • retrieval augmented generation (rag): integrated to provide relevant study content dynamically.
  • memory agent: tracks student performance and automatically adjusts the revision material accordingly.
  • intelligent question bank: ai analyzes past exams and examiner feedback to select the best questions for each student.

entering private beta

by may, markwise was ready for real-world testing:

  • feedback loop: collaborated with actual students, refining features based on direct user feedback.
  • iterative improvements: quickly updated the platform to meet real student needs - from ui tweaks to more complex ai enhancements.
  • early successes: students reported improved revision efficiency and boosted confidence.

markwise doesn’t just improve revision; it makes revision actually effective.


key lessons learned

building markwise taught me a lot, especially about moving fast while staying student-focused:

  • listen to your users: direct student feedback guided nearly every significant improvement.
  • focus on core features: ai-driven personalization was the heart of markwise, and everything else supported that.
  • iterate quickly: small, frequent improvements kept momentum high and users engaged.

what's next?

markwise is still evolving, with the private beta helping refine features further:

  • expanding ai capabilities to cover more subjects and exam boards.
  • enhancing user experience with even smoother interactions and smarter insights.
  • preparing for a broader release later this year.