Skip to main content
All episodes

GILD #9·Nov 16, 2026

Your AI Has Amnesia. Here's How an Amazon PM Fixed It

KS

Kurt Shriner

Amazon Whole Foods

About this episode

In this episode of the GILD Podcast, Gino Ferrand sits down with Kurt Shriner, Senior Technical Product Manager at Amazon Whole Foods, who built his career from help desk to VP of infrastructure at a bank before moving into product at Amazon.

They explore how Kurt went from never writing a line of code to running his day through a team of AI agents, from an adversarial review that now predicts about 90% of his doc-review feedback to a "Memory Steward" handoff system that keeps his AI from starting every session with amnesia.

Drawing from years in infrastructure, enterprise architecture, and IT leadership, Kurt shares why review has become the new bottleneck, where human judgment still has to stay in the loop, and why the real advantage isn't the model, it's the context you give it.

Kurt's views are his own and do not represent Amazon.

Topics include:

■Why AI starts every session with amnesia, and how to fix it

■Building agents and SOPs without ever writing code

■The adversarial review that predicts 90% of doc feedback

■Why review, not production, is the new bottleneck

■Where AI drafts and where human judgment stays

■Why context, not the model, is the real advantage

Chapters:

00:00Your AI has amnesia

00:20Meet Kurt Shriner: "now my team is AI"

01:12Choosing tools: Kiro, MCPs and Claude

02:22The adversarial review that predicts 90% of feedback

04:51"Never wrote a line of code in my life"

05:0630 emails, 2 paragraphs: AI synthesizes, he judges

06:26Do agents work overnight?

07:07Do you still understand what you ship?

07:58"Confident wrong" and building fail-safes

08:47Pitting one AI against another

10:05Which model for which job

11:25Managing cognitive load: watchlist, to-dos, waiting-on

13:59"What's the damage?" Why he reviews every message

15:16Review is the new bottleneck

16:29The 70% rule

17:32Problem statement and SOP before code

18:59How he prioritizes, and the 3-month rule

19:58Memory Steward: a handoff note for your AI

22:01The 3-month memory test

24:04What he'd keep, change and leave out

25:17Why 80% isn't good enough

26:06His AI gets better every week

27:04ChatGPT vs Claude: he wants honesty

28:16"The model is rented, the context is yours"

GILD brings together founders, CEOs, and senior tech leaders to discuss how AI is actually being used and challenged inside companies.

Subscribe for more conversations with leaders building in the age of AI.

Explore GILD

GILD Podcast + more episodes

https://gildhq.com/podcast

LinkedIn

https://www.linkedin.com/company/joingild

Instagram

https://www.instagram.com/gild.hq/

Host

Gino Ferrand

https://www.linkedin.com/in/ginoferrand

Guest

Kurt Shriner

https://www.linkedin.com/in/kurt-shriner/

Presented in partnership with Tecla.

Tecla gives companies flexible access to vetted talent across the US and Latin America, from fractional specialists to dedicated nearshore pods.

Tecla develops its talent for AI-native work, with a focus on automation, agentic systems, and using AI to drive more impact across their role.

Learn more:

https://www.tecla.io?utm_source=spotify&utm_medium=video&utm_campaign=kurt-shriner

#GILDPodcast #AIagents #ProductManagement #HumanInTheLoop #aileadership #futureofwork #founders #techleadership #aiinbusiness

Podcast