PublishedNo. 09
The Return on Compute Era
Controlling the cost of enterprise AI is becoming an architecture problem: what runs on frontier cloud models, what runs on local open weights, and how sensitive data stays inside the building.
Currently GTM at a hyperscaler. Deploying AI solutions at scale. Follow along for those learnings + a monthly running log of personal building, automating, and shipping learnings with AI dev tools — in real time, in public, with all the mistakes left in.
In this issue
Controlling the cost of enterprise AI is becoming an architecture problem: what runs on frontier cloud models, what runs on local open weights, and how sensitive data stays inside the building.
I stopped building assistants a while back. What I run now is closer to a small company with one employee.
Enterprise AI is not waiting for smarter models. It is waiting for models that can reach the work.
Or find me elsewhere