Advanced Vibe Coding Program with Microsoft Certification

Integrate Claude or GPT-4o into a Lovable app via Supabase Edge Functions — with streaming, usage limits, and Stripe.

Topics covered

• AI integration architecture: Supabase Edge Functions as the right layer for AI API calls (never directly from frontend)

• Streaming AI responses: Server-Sent Events in Edge Functions, loading states, thinking indicators, error boundaries

• Usage limits and cost control: per-user token tracking in Supabase DB, soft warning at 80%, hard block at 100%

• Stripe + Lovable: subscription pages in 20 minutes, Free tier (100 credits/month) + Pro tier setup

• Product options: AI Resume Analyser, AI Content Brief Generator, AI Study Tutor, AI Meeting Companion, AI Feedback Analyser, or your own idea

You will learn to

• Add a streaming AI feature to any Lovable app using Supabase Edge Functions

• Implement per-user token limits and usage dashboards in Supabase

• Build a Stripe-powered pricing page with free and pro tiers

Hands-on builds

• Streaming AI Feature build: SSE implementation with loading states and error boundaries

• Project 5 full build day: AI SaaS with auth, core AI feature, usage dashboard, Stripe pricing page

• Final polish: UI, edge cases, loading states, error messages — then deploy to production URL

From builder to strategist: learn how to position, measure, price, and grow an AI product.

Topics covered

• AI product strategy framework: business problem fit, technical feasibility (latency, cost, accuracy), user adoption

• 8 UX principles for AI products: progressive disclosure, confidence communication, graceful failure, human override, streaming, memory

• AI-specific Figma component library: chat bubbles, confidence bars, loading skeletons, correction UI, 'generated by AI' badges

• Miro for AI product managers: user journey maps, agent orchestration diagrams, stakeholder-ready AI product canvas

• AI product metrics and OKRs: task completion rate, tool accuracy, hallucination rate, cost per task, adoption rate, NPS

• 6 AI pricing models: per-seat SaaS, usage-based, outcome-based, freemium + credits, enterprise license, API pass-through markup

• GTM strategy: ICP, '10× better than X' positioning, PLG vs. SLG, AI growth loops, channels (Product Hunt, HN, LinkedIn)

• AI ROI framework: time saved × FTE cost × volume — build a live ROI calculator in Lovable

You will learn to

• Design 3-screen Figma prototype for an AI product using AI-specific UX patterns

• Build a Miro AI product architecture diagram ready for investor or exec presentations

• Calculate and articulate ROI for an AI automation use case

• Build a complete AI GTM strategy with ICP definition, pricing model, and launch channels

Hands-on builds

• Figma for AI Product Design: build 3-screen prototype for capstone using AI-specific components

• Miro Architecture Diagram: capstone end-to-end architecture (RAG + agent + frontend + automation + observability)

• LangSmith + Mixpanel observability dashboard: connect Project 5, track feature usage, conversion funnel

• AI ROI Calculator: Lovable tool — inputs: team size, hours saved, FTE cost → output: annual savings, payback, 3-year NPV

Assignment

Capstone planning: 3-minute pitch of your capstone idea to the cohort. Finalise scope and tech stack.

The ethical foundations every AI builder must understand — and your live pitch to an industry expert panel.

Topics covered

• AI safety fundamentals: hallucination taxonomy, prompt injection (attack + defence), adversarial input testing

• Guardrails AI: input validators, output validators, PII detection, toxic content filtering — integrated into Project 5

• Llama Guard and Nemo Guardrails: open-source safety models for production AI systems

• AI governance and compliance: EU AI Act risk tiers, India DPDP Act, MeitY guidelines, compliance documentation

• AI bias and fairness: training data bias, feedback loop bias, demographic parity, equalized odds, mitigation strategies

• Project 6: 10-slide GTM deck build — Problem, Solution + Architecture, ICP, Figma UX, Tech Stack, Metrics, Pricing, GTM Plan, ROI, Roadmap

• Industry panel pitch: simulated investor/exec Q&A format with 2 guest mentors scoring live

You will learn to

• Add Guardrails AI validators to a production AI application and test adversarial inputs

• Apply the EU AI Act risk framework to classify an AI system and document compliance steps

• Build and present a complete 10-slide AI product GTM deck to a real audience

Hands-on builds

• Guardrails integration in Project 5: test 10 adversarial inputs, document false positive/negative rates

• GTM Deck build: 10 slides with Miro architecture, Figma wireframes, ROI calculator, LangSmith metrics

• Project 6 live pitch: 7-minute presentation + 8-minute Q&A with industry expert panel