Season 5, Episode 38: AMI Vulnerability Scanning, ECS Auto Scaling, and AI Token Costs

Season 5, Episode 38 of the LogiCast AWS News Podcast sees Logicata’s Karl Robinson and Jon Goodall joined by special guest Renato Losio, AWS Data Hero and InfoQ editor, to unpack a fresh selection of AWS announcements and cloud industry news.

From security tooling and container scaling improvements to AI agent extensibility and the commercial future of AI services, there’s plenty to get stuck into. Along the way, the conversation explores both the technical details and the bigger questions surrounding AI adoption, including whether the current pace of growth and investment is truly sustainable.

And, in true LogiCast fashion, Karl, Jon, and Renato somehow spend a solid chunk of the episode debating whether Asda, Tesco, and Hebden Bridge Trousers qualify as acronyms, syllabic abbreviations, or something else entirely. Just another perfectly normal day on an AWS news podcast!

 

06:36 - Amazon Inspector - Direct EC2 AMI Vulnerability Scanning

Amazon Inspector can now scan EC2 AMIs for vulnerabilities without launching an instance. Jon welcomes the simpler security workflow, while Renato highlights the benefits of scanning images before deployment. The group also discusses the ongoing challenge of CVE fatigue and acting on security findings.

 

12:40 - Amazon ECS - Faster Auto Scaling Response Times

AWS has improved ECS auto scaling, reducing response times from around 363 to 86 seconds. Jon welcomes the automatic performance improvements, while Renato questions the choice of a 20-second monitoring interval. Karl suggests the figures may reflect AWS balancing performance and reliability.

 

18:59 - AWS DevOps Agent - MCP Servers, Custom Skills, and Connectors

AWS DevOps Agent can now be extended with MCP servers, custom skills, and connectors to access external operational knowledge. Jon highlights the importance of integrating runbooks and third-party tools, while Renato sees broader applications for agent architectures. The group also discusses the challenges of bringing AI agents into production and demonstrating ROI.

 

28:51 - Amazon Bedrock - Revenue Ambitions and Rising AI Costs

Amazon CEO Andy Jassy suggests Bedrock could eventually rival EC2 in revenue. Jon questions the comparison and highlights rising AI pricing, while Renato points to service quotas and developer experience as adoption barriers. The group discusses increasing token costs and the shift away from fixed-price AI subscriptions.

 

38:22 - Amazon's AI Memo - Mixed Messages on AI Adoption and Engineering Productivity

Amazon's internal AI memo encourages engineers to embrace AI while warning against excessive token consumption. Corey Quinn questions the conflicting messages around AI adoption, productivity measurement, and employee expectations. The article highlights the tension between reducing internal AI costs and AWS's token-based revenue model.