Using AI for Fun, Profit, Ransom: A Practical Guide to the AI Frontier

About This Event

What happens when cutting-edge technology meets everyday life, big business, and digital threats? An evening that dives deep into the three faces of AI — the good, the profitable, and the downright dangerous. 🧠

🔍 Three Dimensions of AI

• 🎨 Fun: See how AI can supercharge creativity, automate daily tasks, and generate everything from avatars to punchy social posts
• 💼 Profit: Learn how companies are using AI to build intelligent products, optimize performance, and unlock new revenue streams — backed by actual examples
• 🛡️ Ransom: Discover how AI can be weaponized — used by bad actors for cyberattacks, data theft, and digital manipulation — and how we can defend against it

Speaker Bio

Konark Modi is Managing Director of Tesseracted Labs GmbH and an Indo-German Young Leader 40 under 40 (2025). With 15+ years in tech leadership at Hubert Burda Media, HolidayCheck Group, Cliqz/Brave Search, and MakeMyTrip, he is a recognized speaker at DEFCON, PyCon UK, and EuroPython, bringing deep expertise in product vision, security, and AI innovation.

📑 Video Chapters

🎨 Fun — Getting Started with AI

• 00:00 — Welcome & Icebreaker
• 01:22 — Hidden malicious instructions in prompts
• 02:25 — Fun vs. scary: what invisible prompts enable
• 03:01 — Konark's background & why this talk matters
• 03:41 — Three themes: Where to begin with AI
• 04:25 — Using GenAI in tools, functions, and new business models
• 05:39 — How people actually use ChatGPT today
• 06:26 — Too many models? How to simplify choices
• 07:27 — Free vs. paid vs. advanced setups
• 09:10 — Demo: OpenRouter and model comparison
• 11:05 — Demo: LM Studio (offline models)
• 13:07 — AI-native tools (Google Sheets, Adobe, Slack…)
• 14:41 — Prompting as a skill for AI-native workflows
• 15:26 — Using LLMs to build dashboards without uploading data

💼 Profit — AI for Business

• 17:20 — Commoditizing ML + Data
• 18:49 — Embeddings, topic modeling, hyperpersonal podcasts
• 20:43 — Community moderation & call center forecasting
• 21:49 — The simple mental model: LLMs + IR systems
• 23:04 — Prompting at scale: reliability, cost, and latency
• 26:00 — Reasoning models vs. cost blow-ups
• 28:10 — Why speed matters: TTFT & tokens/sec
• 30:06 — 3,300 tokens/sec demo and implications
• 32:56 — Recommender system demo powered entirely by LLMs
• 35:41 — Scaling millions of prompts via Google Vertex
• 37:13 — "Imagine unlimited prompts at scale"
• 39:14 — Cost reality: 10M+ prompts/day ≈ $300–400

🛡️ Ransom — AI Security & Risks

• 40:03 — Real-world security failures with AI tools
• 42:28 — Trust without verification: news errors, user overconfidence
• 44:24 — Deepfake conference-call fraud ($25M example)
• 47:48 — Hidden instructions on webpages triggering agent actions
• 48:39 — "Look at the attack vectors this will open up"
• 50:00 — Malicious LLM models injecting scripts into HTML
• 53:00 — The explosion of LLM attack vectors
• 54:10 — Anatomy of an AI application (and its vulnerabilities)
• 56:28 — Closing: "LLMs are not a magic wand"
• 58:11 — Final words

Trailer

Full Presentation

Tagged with

📌 Related GenAI Wednesday Sessions

• How Mindfuel Embeds AI into Every Corner of the Company — Max Koennings
• AI & Entertainment: Revolutionizing Filmmaking — Albert Bozesan
• GenAI: A Business Game-Changer or a Disaster? — Rainer Karcher
• Local AI for Confidential Documents — Mathias Schlolaut

Want more practical AI talks in Munich? Explore upcoming GenAI Wednesday events.

Event Details

• Venue: jambit GmbH, jambit MeetHub, Birketweg 21, 80639 München
• With the kind support of CompuSafe Data Systems AG & jambit GmbH

Back to All Events