Study Go

Study Go is an AI-powered study coach we built on Superhuman Go’s agent platform to help college students juggling multiple classes, assignments, and deadlines learn more effectively from their own course materials. Unlike general chatbots, it grounds explanations in students’ actual lecture notes and slides from Google Drive, generates targeted practice questions with feedback, identifies knowledge gaps, tracks progress through a structured study record, recommends what to review next, and surfaces real deadlines without inventing information. We designed custom agent instructions around a continuous learning cycle (Learn → Practice → Identify Mistakes → Track Progress → Improve) and configured a scheduled trigger for proactive study support. Our biggest challenges were making the agent guide students like a real tutor rather than handing out answers, enforcing strict grounding rules to prevent made-up deadlines or materials, fixing LaTeX rendering by switching to plain-text math, and distinguishing between what was merely discussed and what was actually saved as progress. We’re proud of turning a simple assistant into a structured, approachable learning experience that adapts across subjects and prioritizes understanding over memorization. Along the way, we learned that building a useful agent depends on clear workflows, reliable boundaries, and careful handling of context and memory, and that AI in education should support learning rather than replace it. Next, we plan to add an Exam Rescue Mode, smarter long-term memory, a Teach-Back Mode, calendar integration, and a progress dashboard. Our goal is to make personalized, high-quality academic support accessible to every student.