Session 15 Slides: AI Product Lab & Showcase

--:-- --
↓ Scroll for more

Session 15

AI Product Lab & Showcase

AI Product Engineering

Block 3: Agents, Evaluation & Deployment

Final Session — 30 Hours Complete

Today’s Agenda

  • What Makes a Strong Showcase
  • Presentation Structure (7 minutes)
  • Live Demo Rules
  • What You Have Built in 30 Hours
  • Final Deliverables Checklist
  • THINK → KNOW → BUILD → CONNECT → EQUIP → EVALUATE → SHIP
  • Interactive: Demo Ready? (3–4 min)
  • Interactive: Showcase Gate (3 min)
  • Lab: Prep (0–45 min)
  • Lab: Showcase (45–120 min)

What Makes a Strong Showcase

🎯 Real Problem

Audience understands who has this problem and why it matters

⚡ It Works

Live demo of core AI functionality — not just slides

🧠 Critical Thinking

Honest about what works, what fails, what's next

Presentation Structure (7 minutes)

SectionTimeContent
Problem90sWho, what, why it matters
Solution60sOne clear sentence about what your app does
Live Demo2.5mEnd-to-end with a real example
Architecture60sOne data-flow diagram
Evaluation60sWhat you tested, worked, failed
What Next?30sOne specific credible next step

Live Demo Rules

  • Prepare 3 inputs you know work — do not improvise
  • Show a failure case too — "this doesn't work because…" is impressive
  • Have a fallback: screenshots or recording if API fails
  • Narrate the architecture while demonstrating

Interactive: Demo Ready? (3–4 min) (1/3)

Peer coaches: cold-call a team on each question.

Interactive: Demo Ready? (3–4 min) (2/3)

Peer coaches: cold-call a team on each question.

Interactive: Demo Ready? (3–4 min) (3/3)

Peer coaches: cold-call a team on each question.

What You Have Built in 30 Hours

✅ AI-native development environment

✅ Structured prompt library

✅ Personal knowledge vault

✅ Express application with Gemini on the server

✅ Vector embeddings + RAG pipeline

✅ Multi-step agentic workflow

✅ Tool calling integration

✅ Multimodal data pipeline

✅ Test suite + evaluation report

✅ AI Risk Report

Final Deliverables Checklist

  • Gate: no showcase without test suite scores and the AI Risk Report
  • Working Express app with Gemini on the server (not in the browser)
  • RAG pipeline with ≥10 documents in knowledge base
  • At least one tool implemented
  • Test suite: ≥15 test cases, documented scores
  • AI Risk Report (security, privacy, bias)
  • Architecture diagram
  • Three prepared demo inputs + one honest failure case

Interactive: Showcase Gate (3 min) (1/3)

Which teams are showcase-ready?

Interactive: Showcase Gate (3 min) (2/3)

Which teams are showcase-ready?

Interactive: Showcase Gate (3 min) (3/3)

Which teams are showcase-ready?

Lab: Prep (0–45 min) (1/2)

  • Three known-good demo inputs + one honest failure
  • Fallback screenshots / recording if the API dies

Lab: Prep (0–45 min) (2/2)

  • Architecture: browser → Express → Gemini; vault → embeddings.json → retrieve
  • Gate: scored test-suite.json + AI-Risk-Report.md or no showcase

Lab: Showcase (45–120 min)

7 min + 3 min Q&A per team. Structure:

SectionTime
Problem90s
Solution60s
Live demo2.5m
Architecture60s
Evaluation60s
What next?30s

THINK → KNOW → BUILD → CONNECT → EQUIP → EVALUATE → SHIP

You have passed through the complete AI engineering cycle.

This framework applies to any AI application, at any scale.

You are now an AI product builder.