Lab · AI workflow
My Potential
An academic-guidance product that connects questionnaires to a catalogue of university degrees and vocational programmes, then generates a tailored report with AI and RAG.
- Status
- Working prototype
- Year
- 2025
The problem I saw
Choosing what to study means matching interests and preferences against a large education catalogue. Questionnaires produce a profile; catalogues list options. My Potential connects the two so the resulting guidance conversation can be more specific.
Idea and scope
I built the complete flow: questionnaires, a structured education catalogue, selection of related options, and report generation. Rails and Hotwire run the product; OpenAI and Qdrant handle generation and context retrieval.
What I built
- A questionnaire engine covering interests, traits, and preferences without presenting the results as a diagnosis.
- A catalogue of Spanish university degrees and vocational programmes, structured so each option can be related to a student's answers.
- A pipeline that retrieves catalogue context through Qdrant and uses OpenAI to draft the report from the answers and selected options.
- The full-stack product in Ruby on Rails and Hotwire, including its database and deployment with Docker and Kamal.
- A privacy flow that does not ask for or store a student's name or email address.
Where it stands
The prototype can run the questionnaires and generate a report. The demo is public; screenshots and any claims about questionnaire validation remain out of scope until they can be confirmed.
What it shows
- Design and engineering of a complete product, from questionnaire and education data models through to the final report.
- RAG grounded in a controlled catalogue rather than asking the model to suggest options without source context.
- Privacy decisions applied to both the flow and data model when the product may be used by minors.