Making AI Useful
summary
Company builds on Nvidia hardware and moves focus from training to inference in 2025, claiming 3.5x ROI on GenAI projects and up to 8x for top players, urging migration from old CPU racks to Blackwell GPUs to cut power and space. Real results include legal document processing that reduces review from weeks to minutes or hours, a 53rd-iteration chatbot reaching user preference, and a computer vision system that stops culling male chicks. Storage and networking remain the practical bottlenecks.
AI-written summary · about
chapters (9)
- 0:00Introduction and background
- 2:08AI hype to real impact
- 5:19Legacy hardware and GPU acceleration
- 9:51Why AI must be useful
- 13:23Human-centric AI design
- 15:27Real-world AI use cases
- 20:03Call to action for useful AI
- 21:27Q&A: Storage and networking bottlenecks
- 22:53Q&A: Biggest pitfalls in AI adoption
related talks
03mClosing Remarks
32mRethink business value for generative AI use cases
16mUnleashing the Future of AI Development
38mAI’s Evolving Landscape: Opportunities Strategy Economics and Sustainable Advantage
29mWhen to Gen AI? A strategic framework for real-world AI adoption
53mHow to Build and Scale Ultra-Lean Gen AI Startups
05mOpening Remarks
13mFaster AI time-to-value when planning