DiBotak
DiBotak
traceO: An AI learning engine built around practice
/golangvuetypescriptaipostgressse

traceO: An AI learning engine built around practice

A deployed learning product that turns a topic into a progressive course and makes learners apply the material, built with Go, PostgreSQL, Vue, and an OpenAI-compatible LLM.

Live Demo

traceO — implementation case

Implementation focus: progressive course generation with visible work-in-progress, resumable steps, and practical learner interaction.

Problem

A conventional course can make it easy to consume material without applying it. The product needed to turn an open-ended topic into a structured learning path, then give the learner a reason to do the work rather than only watch or read.

Implementation

traceO transforms a topic into a four-stage blueprint—Hook → Core → Sandbox → Payoff. Hook and Core are prepared up front; Sandbox and Payoff are generated when the learner reaches them. This makes practice part of the product flow rather than an optional extra.

The application uses a Go API with PostgreSQL and a Vue 3 + TypeScript frontend. Generation runs through a visible FIFO worker; stages persist as they complete, and server-sent events stream progress to the interface. Provider configuration is stored in the database, allowing the endpoint/model settings to change without a redeploy.

Decisions and constraints

  • Persist each completed stage so an interrupted build can resume from its partial blueprint.
  • Make topic preview free and editable before queueing generation.
  • Keep queue state visible and cancellable instead of hiding long-running work.
  • Default shared course links to read-only access.

Outcome

The deployed product at traceo.dibotak.com delivers the complete flow from topic preview through generated blueprint and interactive stages. No usage or business-impact metrics are claimed here.

Stack

Go, PostgreSQL 16, Vue 3, Vite, TypeScript, server-sent events, and an OpenAI-compatible LLM endpoint.

Back to Projects