| """ |
| Create India AI Impact Buildathon presentation for ScamShield AI. |
| Uses the official template and fills it with content from the PDF guidelines |
| and project documentation. |
| """ |
| from pathlib import Path |
|
|
| ROOT = Path(__file__).resolve().parent.parent |
| PPT_DIR = ROOT / "PPT" |
| TEMPLATE_PATH = PPT_DIR / "India AI Impact Buildathon - Sample PPT.pptx" |
| OUTPUT_PATH = PPT_DIR / "ScamShield_AI_India_AI_Impact_Buildathon.pptx" |
|
|
| |
| SLIDE_CONTENT = { |
| 1: [ |
| ("INDIA AI IMPACT BUILDATHON", "INDIA AI IMPACT BUILDATHON\nChallenge 2: Agentic Honeypot for Scam Detection"), |
| ], |
| 2: [ |
| ("gjhghjgjhg", "ScamShield AI"), |
| ("[Your Logo Here]", "[Your Logo Here]"), |
| ("Lorem Ipsum is simply dummy text.\nLorem Ipsum is simply dummy text of the printi", |
| "Agentic honeypot system that detects scam messages, engages scammers with AI personas, and extracts actionable intelligence (UPI IDs, bank accounts, phishing links). Built for India AI Impact Buildathon 2026."), |
| ], |
| 3: [ |
| ("<Heading>", "1) The Problem"), |
| ("[ IMAGE ]", "[ IMAGE ]"), |
| ("Lorem Ipsum is simply dummy text of the printing and typesetting industry. Lorem", |
| "What is happening today: 500,000+ scam calls/messages daily in India; Rs 60+ crore daily losses; 47% of Indians affected or know victims. UPI fraud, fake loans, police/bank impersonation are predominant. Why it matters: Existing solutions only detect passively. ScamShield actively engages to extract intelligence while scammers remain unaware they are talking to AI."), |
| ], |
| 4: [ |
| ("<Add your title here>", "2) Our Solution"), |
| ("Lorem Ipsum is simply dummy text of the printing and typesetting industry. Lorem", |
| "We built an autonomous AI honeypot that detects, engages, and extracts."), |
| ("Caption\nSubheading", [ |
| "Detection: IndicBERT + keywords; English, Hindi, Hinglish.", |
| "Engagement: Multi-turn (up to 20) with believable personas.", |
| "Extraction: UPI IDs, bank accounts, IFSC, phone, phishing links.", |
| "API: REST endpoint for Mock Scammer API integration.", |
| ]), |
| ], |
| 5: [ |
| ("<Add your title here>", "3) How It Works (Simple Flow)"), |
| ("Lorem Ipsum is simply dummy text of the printing and typesetting industry. Lorem", |
| "Input: A message (e.g. You won 10 lakh, share OTP). Decision: Is it a scam? If confidence > 0.7, hand off to the honeypot agent. Output: Agent replies in character; we extract UPI/bank/links and return structured JSON. No jargon: message in, scam check, fake victim talks back, we collect proof."), |
| ("Use this space for highlighted text", |
| "Message In -> Scam? -> Engage (persona) -> Extract (UPI/bank/links) -> JSON Out"), |
| ], |
| 6: [ |
| ("<Add your title here>", "4) Proof It Works (15 seconds)"), |
| ("[ IMAGE ]", "[ IMAGE ]"), |
| ("Lorem Ipsum is simply dummy text of the printing and typesetting industry. Lorem", |
| "Demo: POST /honeypot/engage with a scam message; get agent response and extracted intelligence. Test: 90%+ detection accuracy target; 85%+ extraction precision. Real example: Hindi/English scam phrases trigger engagement and structured extraction."), |
| ("Add Pointer here", [ |
| "Live API demo", |
| "Test results", |
| "Extraction proof", |
| "Metrics dashboard", |
| ]), |
| ("01", "01"), |
| ("02", "02"), |
| ("03", "03"), |
| ("04", "04"), |
| ], |
| 7: [ |
| ("<Heading>", "5) A Nuance We Handled"), |
| ("Lorem Ipsum is simply dummy text of the printing and typesetting industry. Lorem", |
| "Mixed language: We explicitly support Hindi, English, and Hinglish (code-mixed) in both detection and agent replies. Over-polite or urgent scam messages: Personas (elderly, eager, confused) are tuned to respond naturally so scammers do not get suspicious. Repeated attempts in one session: Session state in Redis keeps context so we can prolong engagement and extract more across turns."), |
| ], |
| 8: [ |
| ("<Heading>", "6) Trade-Off & 7) Failure Case"), |
| ("Lorem Ipsum is simply dummy text of the printing and typesetting industry. Lorem", |
| "Trade-off: We chose engagement depth over containment (up to 20 turns) to maximize intelligence extraction; we accept slightly higher latency for richer responses. Failure case: Very short conversations may yield no extraction; novel scam templates not seen in training can get lower confidence. We handle these with clear confidence scores and safe fallbacks."), |
| ("Description for\nSample Text", ["Trade-off", "Failure case"]), |
| ("Sample Text", ["Engagement depth vs containment", "Short convos; novel scam templates"]), |
| ], |
| 9: [ |
| ("Submission by:", "Submission by:"), |
| ("Sample text 1\nSample text 2\nSample text 3", |
| "ScamShield AI Team\nIndia AI Impact Buildathon 2026\nChallenge 2: Agentic Honeypot\nContact: missionupskillindia@hclguvi.com\nSubject: ScamShield AI PPT || India AI Impact Buildathon"), |
| ], |
| } |
|
|
|
|
| def get_shape_text(shape): |
| if not hasattr(shape, "text_frame"): |
| return "" |
| return "\n".join(p.text for p in shape.text_frame.paragraphs) |
|
|
|
|
| def replace_in_slide(slide, replacements): |
| """Replace text in slide shapes. new can be str or list of str (assign in order).""" |
| list_indices = {} |
| for shape in slide.shapes: |
| if not hasattr(shape, "text_frame"): |
| continue |
| current = get_shape_text(shape) |
| if not current.strip(): |
| continue |
| for old, new in replacements: |
| if old not in current: |
| continue |
| if isinstance(new, list): |
| idx = list_indices.get(old, 0) |
| if idx >= len(new): |
| continue |
| replacement = new[idx] |
| list_indices[old] = idx + 1 |
| else: |
| replacement = new |
| try: |
| shape.text_frame.clear() |
| shape.text_frame.paragraphs[0].text = replacement |
| except Exception: |
| pass |
| break |
|
|
|
|
| def main(): |
| from pptx import Presentation |
|
|
| if not TEMPLATE_PATH.exists(): |
| raise FileNotFoundError("Template not found: %s" % TEMPLATE_PATH) |
|
|
| prs = Presentation(str(TEMPLATE_PATH)) |
| if len(prs.slides) < 9: |
| raise ValueError("Template has %d slides, expected 9" % len(prs.slides)) |
|
|
| for slide_idx, replacements in SLIDE_CONTENT.items(): |
| slide = prs.slides[slide_idx - 1] |
| replace_in_slide(slide, replacements) |
|
|
| OUTPUT_PATH.parent.mkdir(parents=True, exist_ok=True) |
| prs.save(str(OUTPUT_PATH)) |
| print("Saved: %s" % OUTPUT_PATH) |
|
|
|
|
| if __name__ == "__main__": |
| main() |
|
|
