The one rule that keeps AI forward teams from stalling

Hey, it's Eric.

We’ve all seen it, give a team AI agents and they can move five times faster. Then they stall anyway, waiting on someone else to approve the next step. What’s the point in that?

The fix is a rule we run at Single Grain. We call it command by negation, and it's one sentence:

"I will do X related to Y by Z, unless otherwise instructed."

When someone hits a blocker, they don't wait for an answer. They state what they're going to do, on what, by when, and then they do it. You only step in if you want to change course. Silence means go.

Here's what it sounds like:

"I'll move the approved homepage sections into design for the Q3 launch by Friday noon, unless otherwise instructed. I'll hold the disputed headline and flag it as swappable."

"Client feedback is late, so I'll use the existing approved template for the webinar promo by 4pm, unless otherwise instructed. Final creative can drop in later without blocking QA."

Notice what happened. Nobody asked permission. They set a default, left a window to object, and kept moving. Now you spend your time steering the few things that need it.

This gets more important the more AI you hand your team. An agent can draft the email, build the page, and pull the numbers in minutes. But, if each of those then waits on a human yes, the human becomes the exact bottleneck you were trying to remove. Command by negation takes that bottleneck out. The default becomes forward.

Some people won't want to work this way, and that tells you something useful too. Set the standard and see who runs with it.

Try it this week. The next time someone opens a message with "I was waiting on...", hand them the sentence.

To commanding by negating,

Eric Siu