Current project · pac.university
PAC University
One student app on web, Android and iOS, with LMS, Calendar and Drive built in, and PAC Agent, a Personal Academic Copilot for study help, deadlines and support tickets, also inside ChatGPT.
- Client
- PAC University (pac.university)
- PAC
- Personal Academic Copilot
- Domain
- University student platform
- My role
- Team lead, 2026 to present
The problem
- Too many placesCoursework, calendar and documents each lived in a different tool.FixOne app with LMS, Calendar and Drive built in.
- Slow helpStudy questions, deadlines and technical issues all waited for staff time.FixPAC Agent: study help, deadlines and support tickets, in the app and in ChatGPT.
- Three platforms, one teamWeb, Android and iOS had to launch together with a small team.FixWeb, Android and iOS built together by a team of 4.
- A tight deadlineThe whole platform had to be live in weeks, not months.FixLive in 20 days, with AI-assisted coding across the stack.
A day with PAC
An example day, to show how the pieces fit.
- 09:00Calendar
Today's lectures, in the same app.
- 11:30LMS
Opens the assignment due on Friday.
- 14:00Drive
Lecture notes, one tap away.
- 18:30PAC Agent
Stuck on the assignment. The agent helps with the topic and reminds that it is due Friday.
- 20:00ChatGPT
Asks @PAC for the ticket status, upcoming assignments and deadlines, without opening the app.
How it works
- 01A student asks, in the app or in ChatGPT.
- 02PAC Agent looks up their courses and calendar.
- 03It helps with the work, gives the deadline, or raises a support ticket.
- 04Every run is traced, and changes are evaluated before they ship.
Key insight
AI-assisted coding is a multiplier, not a shortcut. It gave a team of four a 20‑day launch because every change still went through review and tests.
The hard parts
- Speed with quality. Shipping three platforms in 20 days only works if review and tests are never skipped.
- Help the AI can't fake. Students rely on its study help and deadlines, and support tickets are real work for staff, so every answer and action must be traceable.
- The trade-off. AI-assisted coding bought the 20-day timeline; every change still went through review and tests, because speed without them is just faster bugs.
My part
As team lead of four engineers, I led delivery across the web, Android and iOS apps and built PAC Agent, the AI agent for study help, deadlines and support tickets, with tracing and evaluations, plus the real-time chat.
What I took away
- Small teams with AI-assisted coding can ship big platforms, if review stays strict.
- Meet students where they already are: the app, and ChatGPT through MCP.
- Trace and evaluate an agent from day one, not after the first complaint.
Want the detail behind the numbers - architecture, failure modes, what I’d change? Happy to walk through it.
Talk about this project →