TeachSpec
A classroom app where high-school and undergraduate students learn software-product thinking by writing a spec for a real app idea, getting AI-generated code, then critiquing whether the code matches the spec — building the spec-first muscle that the Dev.to discourse says is missing.
high-school CS teachers and intro college CS instructors who want students to think before they prompt
- Lesson packs (30–45 min) that walk students through writing a spec for a calculator, a flashcard app, a study planner — then handing it to an AI
- Side-by-side view of the student's spec vs. the AI-generated code with a 'gap report' showing where the AI improvised
- Teacher dashboard: every student's spec, every gap report, and a class-wide view of where prompts were most often ambiguous
- Free tier for individual students; school licenses at $5/seat/year
Spec-driven development is the explicit lesson the trending Dev.to article lands on — and it's a lesson that almost no CS curriculum actually teaches, even at the intro level.
The cited Dev.to 'Vibe Coding: Endgame' article reportedly has only ~10 reactions — modest reach, not a viral trend. CS-pedagogy discourse around spec-first exists but the addressable market (high-school + intro-college CS teachers) is small and notoriously slow to adopt new tools.Search Results for vibe coding endgame - DEV Community ↗
Could not verify direct competitors via web search (DuckDuckGo returned 403s); spec-first critique workflows are not a saturated category, but general CS-education platforms (Replit, CodeHS, Codio, Codecademy) could easily add a spec-critique feature as a module.Search Results for vibe coding endgame - DEV Community ↗
K-12 and intro-college CS instructors operate on extremely tight per-classroom budgets (often $0–$200/yr); willingness-to-pay for niche pedagogy tools is historically low unless district-mandated or free, dragging monetization below average.Search Results for vibe coding endgame - DEV Community ↗
Spec-driven development is a durable engineering principle tied to LLM hallucination risk, so the underlying pedagogy should remain relevant; however, AI-generated critique of student specs is a fast-moving target that may be absorbed into general-purpose LMS or AI-tutor products.
Technically straightforward — spec editor + LLM code generation + diff/critique UI is a weekend-scale MVP, but reliable grading of student-written specs and rubric consistency across subjects is a non-trivial pedagogical-engineering problem.