Uploaded July 2026 | Updated September 2026, 2 weeks ago
The largest gains from AI-assisted development are not coming from faster code completion. They are coming from engineering teams that redesign how software gets built. In this episode of AppDevANGLE, Paul Nashawaty speaks with Deepak Singh of AWS and Steve Tarcza of Amazon about the widening gap between incremental AI adoption and AI-native software delivery.
The discussion examines how specifications, structured context, custom agents, automated reasoning and asynchronous execution are changing the software development lifecycle. Singh and Tarcza also explain why teams using the same AI tools can produce dramatically different outcomes depending on how they structure work and govern agent execution. As coding agents become capable of running for longer periods with less supervision, developers are increasingly shifting from direct implementation toward intent definition, validation, orchestration and architectural decision-making.
Key Highlights:
• Why tool access does not guarantee developer productivity
• The shift from coding assistance to agentic engineering
• Specifications as the interface between developers and AI agents
• Context management as a strategic engineering capability
• Why trust is essential for scaling autonomous agents
• The role of automated reasoning in software assurance
• How asynchronous agents are changing developer workflows
• What AI-native software delivery looks like in practice
00:00 - Intro
00:07 - Title: "Innovations Unveiled: Balancing Time, People, and Maintenance
02:47 - AI and the Productivity Puzzle: Bridging Gaps and Accelerating Teams
09:14 - Reimagining Collaboration: Context-Driven Agents as Teammates
11:48 - Trust and Transformation: Embracing AI and Spec-Driven Development
16:53 - What’s Next for Kiro
19:31 - Closing Thoughts
The largest gains from AI-assisted development are not coming from faster code completion. They are coming from engineering teams that redesign how software gets built. In this episode of AppDevANGLE, Paul Nashawaty speaks with Deepak Singh of AWS and Steve Tarcza of Amazon about the widening gap between incremental AI adoption and AI-native software delivery.
The discussion examines how specifications, structured context, custom agents, automated reasoning and asynchronous execution are changing the software development lifecycle. Singh and Tarcza also explain why teams using the same AI tools can produce dramatically different outcomes depending on how they structure work and govern agent execution. As coding agents become capable of running for longer periods with less supervision, developers are increasingly shifting from direct implementation toward intent definition, validation, orchestration and architectural decision-making.
Key Highlights:
• Why tool access does not guarantee developer productivity
• The shift from coding assistance to agentic engineering
• Specifications as the interface between developers and AI agents
• Context management as a strategic engineering capability
• Why trust is essential for scaling autonomous agents
• The role of automated reasoning in software assurance
• How asynchronous agents are changing developer workflows
• What AI-native software delivery looks like in practice
00:00 - Intro
00:07 - Title: "Innovations Unveiled: Balancing Time, People, and Maintenance
02:47 - AI and the Productivity Puzzle: Bridging Gaps and Accelerating Teams
09:14 - Reimagining Collaboration: Context-Driven Agents as Teammates
11:48 - Trust and Transformation: Embracing AI and Spec-Driven Development
16:53 - What’s Next for Kiro
19:31 - Closing Thoughts
