Qoollege

Problem

Applying to university is a sprawling, deadline-driven process spread across spreadsheets, counselors, parents, and a dozen disconnected tools. Students struggle to know which schools are realistic, what to do next, and how to stay on track — and the people supporting them have no shared, trustworthy view of progress. Generic advice doesn't account for a student's actual GPA, test scores, interests, or whether they're a first-year or transfer applicant.

Approach

We built Qoollege as a Turborepo monorepo — a Next.js 16 / React 19 webapp on a Neon Postgres database (Drizzle ORM, ~87 tables) plus a Payload CMS marketing site. The heart is 'Ollie', a conversational advisor on the Vercel AI SDK (Gemini 2.5 Pro) with five server-side tools — getRecommendations, getUniversityInfo, getPlatformInfo, getTaskTypes, and proposeTask — so chat turns into concrete actions like a recommendation list or a task card, not just text. A recommendation engine classifies universities Safe / Target / Reach using acceptance-rate logic plus AI enrichment that explains the fit and suggests how to strengthen the application. Students get SAT/ACT practice with scored sessions, an essay/PIQ workspace with draft history, and a taskboard seeded from college-application milestones — first-year and transfer paths handled distinctly. Better Auth drives role-based access for four signup roles (student, guardian, counselor, partner); parent-linking grants read-only visibility into a student's progress. A gamification layer (badges, an append-only XP ledger, streaks) keeps students returning, and Stripe powers free / standard / plus tiers.

Outcome

A single platform where a student's profile — grades, scores, interests, transfer status — drives personalised recommendations, an actionable taskboard, and an AI advisor that can act on the student's behalf, while parents and counselors share one read-only view of progress. It continues as our own product line, shipping features continuously.