TED Tech Everything you need to know about AI agents | Swami Sivasubramanian
22 snips
Mar 6, 2026 
Swami Sivasubramanian, VP of Agentic AI at AWS and builder of services like DynamoDB and SageMaker, explains autonomous AI systems. He defines what agents are, contrasts them with chatbots, and outlines milestones for trust and accessibility. He describes how agents can reshape software development, automate workflows like video recaps, and become invisible collaborators that boost creativity and productivity.
AI Snips
Chapters
Transcript
Episode notes
From 10 Minutes Of Computer Time To Building AWS
- Swami grew up in rural India with only 10 minutes of weekly computer access and became a "human compiler" to maximize that time.
- That constraint sparked his passion for programming, leading to college, a PhD, and work building DynamoDB, SageMaker, and Bedrock at Amazon.
AI Agents Are Autonomous Goal Driven Systems
- AI agents are autonomous software that reason, plan, adapt, and take actions toward user-defined goals rather than just replying to prompts.
- They sense and interact with digital environments, converting high-level objectives into executable steps and improving over time.
Agents Turn Weeks Of Work Into Hours
- Agents can replace lengthy, manual project work by taking a high-level goal and figuring out the implementation steps, like planning experiments or writing code.
- Tasks that once took weeks can be compressed to hours or minutes as agents synthesize results and reflect on failures.