PLAY PODCASTS
Ep183: Agentic AI - From hype to business impact
Episode 183

Ep183: Agentic AI - From hype to business impact

AWS for Software Companies Podcast · Nate Goyer

December 16, 202533m 21s

Audio is streamed directly from the publisher (rss.art19.com) as published in their RSS feed. Play Podcasts does not host this file. Rights-holders can request removal through the copyright & takedown page.

Show Notes

Four enterprise AI leaders from Box, Snorkel AI, Sumo Logic, and Talkdesk peel away the hype and share battle-tested strategies for implementing agentic AI at scale.

Topics Include:

  • Carol Potts introduces panel featuring AI leaders from Box, Snorkel AI, Sumo Logic, and Talkdesk
  • Diego Dugatkin explains Box serves 120,000 enterprise customers with 1.5 exabytes of secure cloud content
  • Kui Jia shares Sumo Logic processes petabytes daily across 10 AWS regions for intelligent operations
  • Yunjing Ma describes Talkdesk's evolution from contact center to customer experience automation through agentic AI
  • Dennis Panos positions Snorkel AI as leader in embedding human knowledge into data-centric applications
  • Diego reveals Box uses AI internally for faster development and externally for metadata extraction automation
  • Kui explains security teams face overwhelming volumes, sometimes 1,000 signals daily, many AI-generated attacks
  • Sumo Logic announces SOC analyst agent in customer beta and query agent in general availability
  • Yunjing details Talkdesk's multi-agent hierarchy architecture powered by unified TalkDesk Data Cloud platform
  • Four key areas identified: discovery of opportunities, building knowledge-powered agents, optimization, and measurement
  • Dennis emphasizes starting with trusted data foundation before adding generative AI capabilities to avoid hallucinations
  • Diego stresses governance importance: AI guardrails plus traditional data security create comprehensive protection framework
  • Kui warns POC-to-production gap requires intentional design: different latency, accuracy, and security requirements at scale
  • Yunjing shares customer success: 80,000 daily calls, 11,000 documents, 97% accuracy despite complex compliance rules
  • Key success factors include prompt engineering optimization and real-time data processing mechanism improvements
  • Diego advises learning AI tools end-to-end: from ideation through functional demos without traditional prototyping delays
  • Dennis recommends robust evaluation frameworks across system components, similar to software unit testing approaches
  • Yunjing reinforces data processing optimization and governance remain essential alongside exciting agentic AI capabilities
  • Kui urges immediate action: technology evolves rapidly, perfect solutions don't exist, customer focus builds trust
  • Final advice centers on treating AI as digital teammate, not replacement, enhancing productivity and creativity
  • Platform partnerships like AWS Bedrock solve heavy lifting, allowing teams to focus on core differentiators


Participants:


See how Amazon Web Services gives you the freedom to migrate, innovate, and scale your software company at https://aws.amazon.com/isv/

Topics

Generative AI#AWSforSoftwareAgentic AIAIcloud computing providersawsAmazon.comcloud servicesAmazoncloud computingcloud serviceBoxtalkdesksumologicsnorkelAI