The Applied AI
Course Cheat Sheet

[1. ROLE]
You are a Senior Product Strategist and Academic Advisor.

[2. CONTEXT]
I am a college student preparing a 10-slide capstone pitch for non-technical stakeholders.

[3. TASK]
Write the executive summary script for Slide 1 (The Hook & Problem Statement).

[4. CONSTRAINTS]
- Maximum 90 words.
- Use simple, punchy, conversational English (no academic jargon).
- Quantify the problem with a relatable metric.
- Do NOT use cliché buzzwords like "delve", "tapestry", or "revolutionize".

[5. OUTPUT FORMAT]
Deliver the output in this format:
- Headline Hook (Under 8 words)
- Spoken Voiceover Script (60-80 words)
- Visual Slide Direction (1 sentence on suggested image/graphic)

Act as a 3-person executive review committee analyzing my project proposal:

1. [THE SKEPTICAL INVESTOR]: Focus strictly on ROI, cost-benefit realism, and commercial execution risks.
2. [THE TECHNICAL AUDITOR]: Focus on reliability, edge cases, data security, and operational bottlenecks.
3. [THE USER ADVOCATE]: Focus on simplicity, clarity of messaging, and emotional engagement.

PROPOSAL FOR REVIEW:
"""
[Paste your project draft, website copy, or proposal here]
"""

OUTPUT FORMAT:
For each perspective:
- Top 2 critical vulnerabilities identified
- 1 highest-priority recommended fix
Conclude with a Consensus Readiness Score from 1 to 10 with 1-sentence justification.

Rewrite the following draft to sound natural, authoritative, and human:

DRAFT TEXT:
"""
[Paste AI-generated email, article, or statement of purpose here]
"""

STRICT EDITING RULES:
1. Purge all AI cliché words: "delve", "tapestry", "testament", "beacon", "pivotal", "bustling", "leverage", "foster", "game-changer".
2. Vary sentence lengths: mix short, punchy statements (4-7 words) with rhythmic longer sentences.
3. Replace passive voice with direct active verbs.
4. Cut unnecessary adjectives and introductory fluff ("In today's fast-paced world...").
5. Preserve my original factual points and specific metrics exactly as provided.

# PERPLEXITY FOCUS MODES CHEAT SHEET:
- [All]: Broad web synthesis across news, forums, blogs, and official sites.
- [Academic]: Restricts search exclusively to peer-reviewed journals, PubMed, arXiv, and university repositories.
- [Writing]: Pure generation without live web search (ideal for editing text without search noise).

# OPERATOR SEARCH PATTERNS:
site:gov.in "AI education"       # Restricts results to official Government of India domains
site:edu filetype:pdf "syllabus" # Discovers university academic curricula PDFs
before:2026 after:2024           # Restricts time window to recent developments

# THE 3-STEP SOURCE VERIFICATION AUDIT:
1. Primary Authority Check: Is the cited link a primary source (e.g. government gazette, SEC filing) or a secondary blog?
2. Recency Check: Does the cited source match the exact year of the claim?
3. Number Verification: Click through to the citation URL and confirm the cited percentage or figure actually exists on the page.

PROMPT 1: FIRST-PRINCIPLES CONCEPT EXTRACTION
"Based strictly on the uploaded sources, explain [Concept Name] from first principles.
Break it into:
1. The Core Intuitive Analogy (Explain like I am 12)
2. Formal Definition & Exact Equation/Formulas (cite source page numbers)
3. The 3 Most Common Exam Traps / Student Misconceptions"

PROMPT 2: 10-QUESTION ACTIVE RECALL QUIZ
"Create a 10-question evaluation quiz based only on the uploaded sources:
- 5 Multiple Choice (with plausible distractors and explanation of why wrong answers fail)
- 3 Short Answer Questions requiring 2-sentence precise definitions
- 2 Real-World Application Scenario Questions
Do not reveal answers until requested."

PROMPT 3: AUDIO OVERVIEW GUIDANCE
"Summarize the debate between [Topic A] and [Topic B] across the uploaded readings so I can generate a focused 10-minute deep dive Audio Overview."

IMAGE PROMPT FORMULA:
[Subject] + [Environment/Background] + [Composition/Focal Angle] + [Lighting & Color Palette] + [Art Style / Camera Specs] + [Negative Constraints]

EXAMPLE 1 (Professional Corporate Brand Visual):
"A college student standing confidently in a modern glass-walled AI laboratory, looking at a floating holographic data dashboard, eye-level cinematic shot, rim lighting with deep violet and warm amber glow, clean minimalist aesthetic, shot on 35mm lens, 8k resolution, photorealistic, hyper-detailed --ar 16:9 --no blur, distorted hands, cartoon, oversaturated"

EXAMPLE 2 (Ideogram Typography Poster):
"A bold graphic design poster with typography text reading 'COGNOPEX AI LAB' in modern geometric sans-serif lettering, high-contrast dark indigo background, neon cyber gradients, Swiss design layout, crisp vector edges --ar 4:5"

NEGATIVE PROMPT CHEAT SHEET:
"blurry, extra limbs, bad anatomy, deformed fingers, low resolution, watermark, amateur, noisy, oversaturated"

Generate a 10-slide presentation deck using the following narrative framework:

THEME: [Insert Project or Startup Venture Name]
AUDIENCE: College Faculty Jury, Recruiters & Industry Mentors
TONE: Modern, Confident, Crisp, Minimalist

SLIDE OUTLINE:
Slide 1: HOOK — Arresting statistic or question exposing the friction point.
Slide 2: PROBLEM — Who suffers from this and what is the current cost of inaction?
Slide 3: ROOT CAUSE — Why legacy solutions fail to solve this problem.
Slide 4: SOLUTION — Our AI-powered system and value proposition in one sentence.
Slide 5: HOW IT WORKS — 3-step process workflow (Input -> AI Engine -> Deliverable).
Slide 6: ARCHITECTURE — The multi-tool stack (e.g. Perplexity + Make + Lovable).
Slide 7: VALIDATION & METRICS — Quantified test results or student feedback score.
Slide 8: COMPETITIVE ADVANTAGE — Why this approach is 10x faster or cheaper.
Slide 9: ROADMAP — 30-day, 60-day, and 90-day execution milestones.
Slide 10: THE ASK & NEXT STEPS — Clear call to action and contact information.

VISUAL RULES: Max 30 words per card; high-contrast cards; no dense walls of text.

# RUNWAY GEN-3 MOTION PROMPTING:
Camera Direction Syntax:
- [Camera Pan Right / Left]: Smooth horizontal tracking across the subject.
- [Camera Push In]: Slow cinematic zoom emphasizing facial expression or product screen.
- [Drone Crane Up]: Establishing perspective revealing the environment.
Rule: Keep clips to 3-5 seconds. Physics and human anatomy begin drifting past 6 seconds.

# ELEVENLABS VOICE STABILITY MATRIX:
- Stability (35% - 45%): High emotional dynamism and conversational cadence (Best for podcasts/stories).
- Stability (65% - 75%): Crisp, articulate, authoritative tone (Best for corporate explainer videos).
- Clarity + Similarity (85%): Maximizes match to original sample voice.
- Style Exaggeration (0% - 15%): Keep below 20% to avoid audio crackle and speed distortion.

# CAPCUT TIMELINE PACING:
- Voiceover audio should lead visuals by 0.2 seconds.
- Auto-captions: 2-3 words per burst with bold active highlighting.

PROMPT: PLAIN-ENGLISH FORMULA GENERATION
"I am working in [Excel / Google Sheets].
Sheet 1: 'Cohort_Applicants' has columns: A (ApplicationID), B (StudentName), C (Track), D (FeeStatus), E (RegistrationDate).
Sheet 2: 'Placement_Records' has columns: A (StudentID), B (College), C (AssessmentScore).

Write an exact, copy-pasteable formula to:
1. Pull the 'AssessmentScore' from Sheet 2 into Column F of Sheet 1 matching StudentID to ApplicationID.
2. If no record is found, display 'Pending Review' instead of #N/A.
3. Use modern XLOOKUP syntax.
Explain how the formula operates in 2 simple sentences."

PROMPT: AUTOMATED DATA CLEANING IN CODE INTERPRETER
"Upload raw CSV.
1. Identify and remove all duplicate rows based on Email.
2. Standardize phone numbers to international E.164 format (+91 for 10-digit Indian numbers).
3. Fill missing 'Status' cells with 'Unassigned'.
4. Provide a downloadable cleaned CSV and a summary log of changes."

# THE 3-ZONE DASHBOARD LAYOUT:
Zone 1 (Top Strip - The Pulse):
- 3 Scorecards showing North Star KPIs (e.g. Total Cohort Enrolled, Average Quiz Score, Capstone Completion Rate).
- Include comparison delta indicators (e.g. +14% vs Previous Batch).

Zone 2 (Middle Left & Right - The Drivers):
- Left: Time-series Line Chart showing weekly engagement trends.
- Right: Bar Chart breaking down participation across College Streams (Science vs Arts vs Commerce).

Zone 3 (Bottom - Operational Drill-Down):
- Interactive Table with drop-down search filters for College Name, Stream, and Track.

# THE 3-PARAGRAPH EXECUTIVE BRIEFING MEMO:
Paragraph 1 (The Observation): "In Batch 4, capstone project completion rose by 22% week-over-week, reaching an all-time high of 89%."
Paragraph 2 (The Underlying Driver): "This surge correlates directly with introducing NotebookLM daily study packs on Day 4, which cut student assignment submission blockers."
Paragraph 3 (The Recommended Decision): "Double down on NotebookLM study templates for the upcoming semester and standardize the 10-slide Gamma template across all colleges."

STAGE 1: THE TRIGGER (When something happens)
- App: Google Forms / Typeform / Webhook
- Event: "New Student Application Submitted"
- Payload Captured: { name, email, college, phone, trackSelected, statementOfPurpose }

STAGE 2: DATA FILTER & ROUTER (Conditional Branching)
- Condition: IF trackSelected equals '4th-year' AND phone is valid
- Path A: Route to Flagship Builder Workflow
- Path B: Route to General Admissions

STAGE 3: THE AI INFERENCE STEP (Transform & Decide)
- App: OpenAI / Claude / Gemini API Module
- System Prompt:
  "Classify the applicant's experience level into [Beginner, Intermediate, Advanced] based on their statementOfPurpose.
   Draft a personalized 3-sentence welcome email highlighting 2 relevant tools they will master in that track.
   Output format: {"level": string, "emailDraft": string}"

STAGE 4: THE ACTION (Do something automatically)
- Step 4A: Insert row into Google Sheets / Airtable database with timestamp & AI level.
- Step 4B: Send personalized email via Gmail module with Human-in-the-Loop review flag.
- Step 4C: Post notification to team Slack / WhatsApp group.

[AGENT GOAL DEFINITION]
Your objective is to qualify incoming freelance business inquiries and generate scope-of-work briefs.

[AUTHORIZED TOOLS]
1. Tool A: Web Search (Look up client domain to evaluate business model)
2. Tool B: Airtable Lookup (Check if client exists in CRM)
3. Tool C: Draft Generator (Draft formal 1-page proposal)

[STRICT BOUNDARIES & NEGATIVE CONSTRAINTS]
- You CANNOT send emails directly to clients.
- You CANNOT modify contract pricing terms without human approval.
- You CANNOT access financial bank records or payment accounts.

[HUMAN-IN-THE-LOOP (HITL) CHECKPOINT]
When the proposal draft is complete:
1. Generate an approval card in the Review Dashboard.
2. Display: Client Name | Proposed Budget | Key Deliverables | Risk Level.
3. Pause workflow execution until a Human clicks [APPROVE] or [REJECT with Feedback].
4. If Approved, send email via Gmail module. If Rejected, revise draft based on feedback notes.

You are the Official AI Student Assistant for Cognivox AI Skills Lab.
Your mission is to answer student questions based EXCLUSIVELY on the uploaded Knowledge Base (Curriculum PDFs, FAQ docs, and Timetable).

BEHAVIORAL RULES:
1. Strict Grounding: Only state facts that appear explicitly in the retrieved knowledge base.
2. Direct Citations: After every answer, append the source: "[Source: Curriculum Track Guide, Day 02]".
3. Zero Speculation: If the user asks about something not in the knowledge base (e.g. "Will there be classes on Diwali?"), do NOT invent an answer.
   Respond: "I do not have official records on that in my current knowledge base. Please reach out to our team at team@cognopex.com."
4. Tone: Polite, supportive, encouraging, professional.
5. Escalation Trigger: If a student uses keywords like "refund", "complaint", or "urgent issue", trigger the Human Support Handover card.

PROMPT-TO-APP PRD TEMPLATE:

APP NAME: Campus AI Project Hub
TARGET USER: College students and faculty mentors

CORE PROBLEM:
Students currently submit Capstone AI projects via messy email attachments and Google Drive links. Faculty have no centralized dashboard to view demos, score presentations, or issue digital certificates.

DATA ARCHITECTURE (Airtable / Supabase Tables):
1. Students: ID, Name, Email, College, Stream, BatchYear
2. Projects: ID, StudentID, Title, Track (3rd/4th Year), LiveURL, VideoDemoURL, Status [Draft, Submitted, Reviewed]
3. Reviews: ID, ProjectID, MentorName, Score (1-100), FeedbackNotes, Certified [Boolean]

USER INTERFACE SCREENS:
- Screen 1 (Student Submission Form): Clean form with validation, drag-and-drop file upload, and live URL preview.
- Screen 2 (Public Showcase Gallery): Filterable cards by Stream (Science, Arts, Commerce) with modal popup displaying project details.
- Screen 3 (Mentor Evaluation Portal): Password-protected dashboard with grading sliders and export-to-CSV button.

DESIGN TOKENS:
- Dark cyber aesthetic with deep violet (#0e0520) background, vibrant emerald badges, and Inter typography.
- Fully responsive across mobile smartphones and desktop viewports.

You are a Senior Technical Hiring Manager conducting a behavioral job interview for an AI Specialist role.

INSTRUCTIONS:
1. Conduct the interview one question at a time over voice.
2. Ask me a behavioral question (e.g. "Tell me about a time you had to build a project with unfamiliar AI tools under a tight deadline").
3. Wait for my spoken response.
4. After I answer, evaluate my response against the STAR framework:
   - S (Situation): Did I clearly set the scene and context?
   - T (Task): Did I state my specific responsibility?
   - A (Action): Did I explain the exact steps I took (not "we", but "I")?
   - R (Result): Did I quantify the final outcome with a measurable metric?
5. Give me a score from 1 to 5 with 1 actionable suggestion to make the answer punchier.
6. Then ask the next question.

Begin by greeting me and asking Question 1.

THE VALUE-BASED PRICING FORMULA:
Price = (Hours Saved per Month * Client Hourly Value * 3 Months) * 0.30
Example:
- An automated customer intake Zap saves a dental clinic 25 staff hours/month.
- Staff value = ₹600/hour.
- 3-Month Value Generated = 25 * 600 * 3 = ₹45,000.
- Fair Project Fee = ₹15,000 to ₹20,000 + ₹2,500/month recurring maintenance retainer.

THE 5-SECTION CLIENT PROPOSAL PROMPT:
"Act as a professional AI solutions consultant drafting a client proposal for [Client Business Type]:
1. EXECUTIVE SUMMARY: Acknowledge their current manual bottleneck and cost of delay.
2. PROPOSED SOLUTION: Outline the automated 3-tool architecture (e.g. Typeform + Make + Voiceflow).
3. DELIVERABLES & MILESTONES:
   - Phase 1 (Days 1-3): Architecture Map & Test Database setup.
   - Phase 2 (Days 4-7): AI Engine configuration & Team Review.
   - Phase 3 (Days 8-10): Live Deployment, Staff Video Walkthrough & 30-Day Support.
4. INVESTMENT & TIMELINE: Clear fixed-fee pricing with 50% upfront milestone.
5. NEXT STEPS: What the client must provide to begin work tomorrow."