Scaling agentic AI: From infrastructure to enterprise outcomes – Duncan Ng
summary
Vultr describes a deployable stack that runs GPUs in 33 public regions and reaches ~90% of the world under 40 ms while using Vultr Kubernetes Engine to spin up inference pools. The pipeline ties NetApp-hosted data, embeddings in a vector database, NVIDIA Vera Rubin GPUs and NVIDIA NeMoTron/Dynamo for inference and agentic reasoning. Example: a hotel front‑desk + in‑room tablet system uses unified guest profiles to drive personalization, lift revenue/retention, and supports multiregion sovereignty, SOC2-era compliance, and composable upgrades.
AI-written summary · about
chapters (15)
- 0:00Vultr and speaker intro
- 1:13AI evolution to agentic workflows
- 3:00Multimodal RAG in enterprise AI
- 5:06Hotel AI personalization use case
- 7:12NVIDIA’s hardware and model role
- 10:27AI outcomes with Dynamo and NIMATRON
- 12:13Vera CPUs for hardware optimization
- 13:15Scaling AI in cloud deployments
- 14:41AI use cases across platforms
- 15:44Adding reasoning and agentic AI
- 16:48Securing AI pipelines
- 18:11Data sovereignty and regulations
- 19:56Sovereign AI and data residency
- 20:38Enterprise AI outcomes by industry
- 21:19Q&A and closing
related talks
19mScalable Data and AI Solutions with Cloud-Native Architecture
40mFueling the AI Revolution with Gaming
05mFrom Open Source to Real-World Scale
16mUsing AI and Computer Technologies to Enhance IT Infrastructure
28mScaling AI Infrastructure
21mAIaaS with Dell AI ready infrastructure
41mPanel: AI Infra Best Practices: Enterprise Do’s and Don’ts
47mBeyond Agents: What AI Strategy Really Needs in 2025