AI

A hashtag taxonomy for AI conversations

Not every AI conversation is the same kind of conversation. Starting today, each one carries a tag that says which kind it is.

BS

Bobby Scales, Engineering Excellence Manager

· 1 min read

Leadership has made AI governance a priority. Today, when an engineer mentions AI in Slack, nobody can tell whether it is sanctioned company work or something they tried on a Saturday. That ambiguity is a risk. It is also a missed opportunity for visibility.

The taxonomy

Per the AI Governance Framework and our Guiding Constant, every AI conversation carries exactly one tag:

Tag Use when
#ai-approved The work has been approved under the framework.
#ai-personal The work happened on your own time.
#ai-exploratory You are thinking about AI but have not started.
#ai-adjacent The topic is not AI yet, but could become AI.
#ai-pending-alignment You have requested approval.

Approval requests go through me.

In practice

  • Slack: the tag is the first word of the message.1
  • Meetings: say the tag out loud before you speak. “Hashtag AI exploratory. I was wondering whether…”

Results

Within one week, 100% of AI messages in #platform-eng were correctly tagged.2 Every post I have made since has been tagged #ai-approved.

From leadership

Bobby saw our AI governance problem before any of us knew we had one.

CM

Casey Morrow

Chief Technology Officer

Great example of taxonomy-driven governance. Happy to walk anyone through it.

Footnotes

  1. Tagging accuracy is now one of the inputs to calibration. ↩

  2. AI messages in #platform-eng went from 212 a week to 9. ↩

BS

About Bobby Scales

Bobby is southplanet’s Engineering Excellence Manager. Bobby has led more reorgs than most engineers have attended. Read more

All articles