✦ From the desk
Acollectionofideas,commentary,andinnovativetakes.
Perspectives on AI, venture, product, and building companies that last.
Why Most Enterprise AI Fails: It's Not the Model, It's the Deployment
OpenAI grew its forward-deployed engineering team from 2 to 52 in a single year. The teams winning enterprise AI aren't shipping better models, they've changed how AI gets deployed inside real organisations.
The Venture Studio of 2030
By 2030, venture studios could sit at the center of how ideas get discovered, validated, and scaled, with design, AI, and modular systems at the core of how companies get built.
What a $500M Ethernet Switch Company Tells Builders About the Next Two Years
Every AI infrastructure cycle follows the same arc: a bottleneck gets solved, then a new layer of applications becomes viable. A $500M Ethernet switch raise is the clearest signal of where the next window opens.
Stop Finding Gaps. Start Finding Pain.
In 2025, AI was the fastest-growing app category, expanding 181% in a single year. In that flood, most builders look for gaps. The ones who matter look for pain.
Ventura, Your Diligence Co-Pilot
Venture diligence hasn't evolved in 20 years, still manual, fragmented, and unintelligent. Ventura is our co-pilot for deal evaluation. This is Project 001.
We Wanted Intelligence, We Got Automation
Enterprises imagined adaptive, self-improving digital colleagues. Most 2025 agent deployments turned out static and event-based. Where we missed the mark, and what's being fixed.
"Some of our favourite projects are still ahead of us."