This is either incredibly stupid or delightfully early, but I’ve been wondering if process control is a necessary pre-requisite for plugging AI into company operations.

Amazon went through a similar journey with inventory reordering (which is 100% automated now).

From Working Backwards:  

The basic idea is:  

1. Do the process using humans 
2. Bring it under statistical control
3. Continue improving — do experiments, trial and error, etc
4. Automate

Writes http://commoncog.com. Books, protocols, and the art of business, from the perspective of an operator. Yes cats.

Seems legit. Assumption that one can throw compute at a vague problem and end up with a scalable, profitable solution is way off. Also—0th step: conceive of and enumerate process?

In motion - complex systems, computing and communication (amongst other things) | msweet.net

The 0th step is the hard bit, and where all the process control shit lives 😄

😂 increasingly looking like I’ll end up digging into your becoming data-driven series. Related: been reading about effectuation (theory of entrepreneurial expertise) and wondering whether 0th step vs steps 1-4 are what separate entrepreneurship and professional management.

😂 increasingly looking like I’ll end up digging into your becoming data-driven series. Related: been reading about effectuation (theory of entrepreneurial expertise) and wondering whether 0th step vs steps 1-4 are what separate entrepreneurship and professional management. https://effectuation.org/effectuation-101