My guest today is Tyler Wells, co-founder of Brain Grid.

Tyler recounts 25+ years in software, from an early IBM XT to work across military communications, startups, Skype/Microsoft, and seven and a half years at Twilio building video and SRE organizations, before founding Propel Data (which didn’t find product-market fit) and then Brain Grid.

He describes an experiment-driven approach to building high-performance systems by defining hypotheses, creating a “steel thread” MVP, and prioritizing observability for 2:00 AM incidents.

He discusses how AI coding shifts focus from typing code to architecture, documentation, critical thinking, and red-teaming plans, while warning that agents need guidance on separation of concerns and DRY to avoid refactor side effects. Brain Grid emerged from using Cursor agents during Propel’s wind-down and aims to generate detailed specs, acceptance criteria, and validation loops so agents implement features reliably, with attention to token efficiency.

He also covers co-founder traits, chaos engineering, compliance challenges for solopreneurs, career advice, and staying grounded through exercise, cooking, and family.

Tyler Wells is the Co-founder and CTO at BrainGrid, BrainGrid is one of the first platforms built specifically to replace the missing product management role in AI-native software development.

He is currently building BrainGrid — helping engineering teams ship faster with AI-assisted requirements breakdown and task management. We're focused on bridging the gap between product ideas and implementation-ready work.

His Background: He has spent 25+ years building systems where failure isn't an option—from satellite communications at Hughes Space to real-time video at global scale. I led the team that built Facebook's first video calling feature powered by Skype, then spent 7+ years at Twilio building their Video Platform (WebRTC) and leading SRE/Observability across the company.