AIWA "The One Thing" #17: Stop Asking "What Is Our AI Strategy?" Start Asking What Ray Dalio Did In 2016.

The sharper question: “Which of your company’s most important processes are closed loops today? Which are still open? How do we close the gaps?”

If you want to close the AI execution gap, think in terms of Open vs Closed loops.

Every important process in your company is a loop.

Closed loops learn. Open loops forget because there are holes that prevent complete information from feeding back in.

The companies that compound are the ones whose loops close.

Ray Dalio was building closed loops at Bridgewater Associates a decade before most leaders had even started considering them.

Recording almost every meeting. Letting employees rate each other in real time. Then feeding all of it back into the system so the firm could “identify and solve problems better” via an intelligence layer.

Closed loop, 2016 edition via CNBC 12.22.16.

Andrew Chen's tweet below caught my attention.

You aren’t going to get an argument from me that taste is a big deal but the bottleneck he’s describing is downstream of something bigger.

The PM bottleneck is one visible ripple. The same architectural shift is showing up in Sales, Marketing, Finance, Operations and beyond.

Artificial Intelligence is becoming the Operating System your company runs on.

Every important action should produce an artifact the system can learn from. The more it learns, the more useful it gets. The asset gets stronger over time.

s / o Ben Lang of Cursor for the Y Combinator video notes

A closed loop can be thought of as an end-to-end process designed so that each component part is measurable and something the system can learn from (i.e. a call transcript).

What worked? What didn’t? Why? Every cycle teaches the system something. Every lesson makes the OS smarter.

I wrote about this in AIWA “The One Thing” #01 as Context Graphs.

Same idea, different lens:

“A Context Graph captures the 'why' behind business decisions by turning institutional knowledge that lives in emails, Slack threads, decks and senior leaders’ heads into searchable organizational memory that AI agents can learn from and apply, in real time.”

Here’s what this looks like in practice.

Within Sales and Marketing, a renewal agent proposes a discount, pulls the supporting context from across systems, routes the exception and records why the call was made. Next quarter, the system already knows.

s/o Jaya Gupta, Ashu Garg, Foundation Capital

“Automate. Augment. Rewire.” is the playbook. Closed loops are what you’re building toward.

The verbs make sense once you see the noun they’re building.

Leadership can now move past the abstract version of the conversation, the jazz hands and confusion that typically follows.

*"Boards + CEOs Don't Agree on Who Owns AI Strategy" via Fortune

Instead, they can identify and stack-rank the open loops. Then convert them to closed loops aka assets that compound.

BIG s/o to Tatiana Ferreira

Humans + Machines. Never Humans vs. Machines.