Ep231: Lead with AI on AWS
Notes de l’épisode
See how enterprises run AI in production on AWS, from frontier models on Amazon Bedrock to Amazon Bedrock AgentCore for governing, securing and tracing agents across every system
Topics Include:
- Why software breaks into five layers: persona, interface, logic, platform, data.
- AI is changing who uses software and how they interact with it.
- Workflows are breaking out of single SaaS tools and running across multiple providers.
- Zero-copy data access unlocks insights trapped in separate departments.
- One AI surface replaces three logins across Salesforce, SAP and Adobe.
- Governing agents: trusted registries, cost controls and token management.
- Security has to run at machine speed when vulnerabilities surface daily.
- Tracing and evaluating workflows that hop between different systems.
- Margin compression: what token costs do to the economics of AI products.
- Four pricing models, and why seats-plus-tokens hybrids are winning.
- Embed, extend, expand or create: a two-by-two grid for AI products.
- Real examples: Intercom Fin, Slack AI tiers, Workday HiredScore.
- First mover or fast follower? AI-built PR FAQs speed up product decisions.
- Why roughly 70% of developer time isn’t spent coding.
- MCP as the new TCP/IP for multi-vendor agent ecosystems.
- Why non-technical salespeople are quickly embracing AI interfaces.
- The four layers: AI surface, harness, reasoning models and chips.
- Why context and memory in the harness make or break agents.
- Building for security, scale and open standards across the AI stack.
- Will models eat the harness? Data and context take centre stage.
Participants:
- Mahdi Sajjadpour – Director, Solutions Architecture, Strategic ISVs – Amazon Web Services
See how Amazon Web Services gives you the freedom to migrate, innovate, and scale your software company at https://aws.amazon.com/isv/