LLMs - application and impact on business
summary
Company builds hardware‑agnostic visual‑language models that run without GPUs for national projects in low‑resource languages, collecting spoken and visual corpora rather than only web text. Deployments cut integration complexity and save about 30% in development budgets by replacing multi‑stage pipelines with end‑to‑end models; corporate search pilots show ~20% lower costs. Engineering focuses on model distillation up to 100x and edge/federated architectures because training at scale needs 100k–1M+ GPUs for next‑gen multimodal models.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 1:51LLMs development journey
- 10:58LLMs revolutionizing business
- 17:10LLM integration strategies
- 22:33Future challenges
- 29:21AI productivity and collaboration
- 35:52Next AI breakthroughs
- 37:40Book recommendations
related talks
15mLLMs for Data-Driven Decision Making in Product Development
25mvLLM: Easy, Fast, and Cheap LLM Serving for Everyone
49mA Practical Approach to Using LLMs: Peculiarities and Pitfalls
01h 19mBuilding an LLM on your Laptop to...
10mThe Impact of LLMs on the Circular Economy and Beyond
16mLLMs are not the product
47mOwn Your LLM: Running Generative and Agentic AI Locally
36mLLMs and AI agents demystified