What we talked about
Kyle Bowman, Senior Technical Product Manager at Amazon leading the Climate Pledge-Friendly Program, joins host Federico Ramallo on the PreVetted Podcast for a wide-ranging conversation that starts with problem-solving in unfamiliar industries and quickly evolves into deeper topics.
Show notes
Kyle Bowman spent four hours trying to get an AI-generated script to download a single table from a document, and that experience became a precise illustration of why starting with the solution rather than the problem is the defining mistake of the current technology moment. The conversation ranges from vibe coding to hyperinflation to Argentinian labor union reform, connected by a consistent thread: how you measure whether you’ve actually solved anything.
What we covered
- Kyle’s career has moved through industrial engineering consulting, Amazon Robotics, international aviation supply chains, and now software product management, and he argues the unifying skill across all of them is defining the problem before reaching for a solution. He points to the Six Sigma “five whys” framework as one useful structure, but emphasizes that the methodology matters less than getting a crisp problem definition first.
- On GenAI, an MIT study he cited found that 98% of enterprise GenAI integrations over a recent three-month period failed to generate revenue. His diagnosis: companies are throwing a solution at everything without first identifying which problem it actually solves, the same trap as “vibe coding” a website nobody will visit.
- He recounts a real attempt to vibe-code a simple Quip-to-Word conversion script: it worked for basic text but couldn’t handle images or tables, and he spent four hours trying to fix one table. His point is not that AI tools are useless but that someone with engineering fundamentals would have spotted the structural problems in the generated code immediately, and that core skills become scarcer and more valuable as solutions become cheaper to produce.
- A security researcher demonstrated how an AI model could be poisoned by seeding GitHub with repositories containing backdoored code and a description that matched common security problem patterns, when the model retrained on GitHub, it learned to recommend the compromised code. Kyle called it “social engineering, but with large language models.”
- The conversation moves into economics through the lens of Argentina’s hyperinflation, 70,000% projected annual inflation in late 2023, with prices changing between morning and evening on the same day. Kyle connects Javier Milei’s approach to the core product management discipline of picking one metric, putting in tough solutions, and watching whether the number actually moves.
- On free markets and monopoly, Kyle observes that competitive markets naturally trend toward monopoly as winners keep winning, which is why the debate isn’t really about free markets versus regulation, but about where to draw the line. He and Federico trace this from 19th-century mining company violence to Argentina’s labor unions conducting 9,700 strikes in 2022.
About Kyle
Kyle Bowman is a Senior Technical Product Manager at Amazon, where he leads the Climate Pledge-Friendly Program, focusing on making sustainable shopping more accessible through product standards and AI/ML-powered discovery of environmentally friendly products.
- LinkedIn: http://www.linkedin.com/in/kyle-bowman-49125544
- Website: https://amazon.com
Episode 82 of the PreVetted Podcast.