CopaMigo
AI Strategy · Student Success Tools · Case study
A multilingual AI triage tool that answers in the student’s own language and routes their plain-language problem to the right human service with a warm handoff card.

Goal
Get a stuck student to the right person quickly, without making them name the office or know the vocabulary, at the moment that often decides whether they stay enrolled. Every successful handoff is a small persistence intervention.
Audience
Students hitting life or academic barriers, especially first-generation and Spanish-speaking students, who are not sure where to go and are often the least likely to walk in and ask. CopaMigo is stigma-aware: it meets them in their own words first.
Process
A student types or speaks their situation and CopaMigo reads it for urgency and need. It then either answers directly from a curated answer bank, written for the question rather than retrieved from the website, or points them to the right office with a warm handoff card showing the contact, hours, and what to ask for. It is anonymous by design, no login and no student record. The full routing model, crisis handling, accessibility, and rollout plan live in the PRD.
Technology
- A single self-contained HTML file: CSS and vanilla JavaScript, no framework, no backend.
- Ships two ways from one codebase: a standalone page and an embeddable widget.
- Multilingual by design, with voice in and out via the Web Speech API.
- More than 100 verified GCC service links and a plain-JavaScript program checksheet lookup.
- The Claude API. A server-side proxy is planned for the wider pilot.
- Hosted on GitHub Pages, mobile-first.
Outcomes
In pilot in the Digital Media Arts program’s Discord, where students are testing it. It was tested first with students in the Design Studio course, and the routing and the answer bank were revised from what those sessions surfaced. No outcome numbers are claimed yet; the next phase, inside the Canvas Student Resources button, is where those get measured: whether a student reaches the right office faster and with less friction than searching the college website, whether the questions students actually ask line up with the answers each office verifies, and where answers are missing so the office that owns the service can write them.
The answers are hand-written from the questions students bring me, in consultation with the offices that handle each service, and have not yet been verified office by office. The staff knowledge builder is built for that step and not yet run, and getting every department through it is a prerequisite for production. The curated questions and answers, plain-language routing, and service links work on their own and can also seed an enterprise platform.