Launching Solveit, the Antidote to AI Fatigue Author: Johno Whitaker Published: October 2, 2025 Categories: Education, Coding, AI --- Overview Solveit is a new platform and approach designed to help programmers and learners avoid the overwhelm of AI-generated code they don't understand. After a successful year-long preview with 1,000 users, it’s now open for signups until October 20th, 2025. The platform supports a wide range of activities including programming, web development, AI projects, system administration, writing, and business. --- Background and Origins Solveit was born from a course titled "How To Solve It With Code", co-created by co-founder Jeremy Howard and Johno Whitaker. The approach builds on the problem-solving framework from George Polya’s 1945 book "How to Solve It": Understand the problem Devise a plan Carry out the plan Reflect and learn These heuristics extend beyond math to coding, writing, learning, and more. --- Key Principles of Solveit Approach Avoid Blind AI Reliance Mindlessly accepting AI-written code leads to: Lack of foundational knowledge Growing technical debt Limited ability to solve complex or novel problems Instead, embrace iterative learning and skill development alongside AI. Iterative and Exploratory Coding Solveit encourages breaking problems into small manageable steps with live feedback. Example: solving an Advent of Code problem by writing small functions incrementally while verifying outputs. This leads to faster comprehension and improved problem-solving skills. Fast Feedback Loops Programming in notebook-like environments grants near-instant feedback on code snippets. Fast iteration reduces mistakes and facilitates exploring alternatives promptly. Fast feedback is critical to learning and innovation, echoed by lean startup principles and data science competitions like Kaggle. --- AI Integration in Solveit Shared Context: AI sees exactly what humans see, including code, notes, variables, and tests, enabling deeper collaboration. Dialog Engineering: Unlike typical chat models prone to compounding errors, Solveit allows editing or removing past AI responses and controlling AI memory, which maintains relevance and effectiveness over longer sessions. Python functions can be used by AI as tools to fetch additional info or execute tasks, enhancing AI usability. --- Design for Collaboration, Not Replacement AI outputs code but does not automatically run or merge it, encouraging users to review and understand first. ‘Learning’ mode encourages generating small steps of code rather than large chunks. AI autocomplete is disabled by default to promote active engagement. Editor design supports dialog and incremental building rather than large code dumps. --- Learning Mindset & Mastery Solveit emphasizes a growth mindset: stop to explore and understand new concepts AI introduces rather than skipping them. This deliberate practice is likened to exercising or practicing an instrument. Evidence shows participants in the first cohort produced hundreds of impressive projects. Continuous learning alongside improving AI capabilities fosters accelerated progress. --- Invitation to Join Solveit offers a five-week course starting October 20th, featuring lessons from founders and special sessions with Eric Ries on impactful startup building. Early signups gain access to the platform and the approach used by the Answer.AI team daily. Signups are open at https://solve.it.com until October 20th or until capacity is reached. --- Summary Solveit is a thoughtfully designed platform that combines classic problem-solving heuristics with modern AI-assisted coding and exploration tools. It prioritizes human understanding, iterative development, and collaboration with AI to deepen skills and avoid the pitfalls of