3 Comments
User's avatar
Mark Palmer's avatar

Good post, but I'm left wondering: how do you suggest this authority be discovered, cataloged, maintained, and applied at runtime during prompting and by AI agents? Any recommendations about how to "manage" "authority?" How do you map Enterprise Identity to Enterprise Authority?

Expand full comment
Jayson Winchester's avatar

Good question. I don’t think this requires boiling the ocean. Start where risk lives: the actions that bind the org. Model proposes; system checks a mandate exists for that commitment class and limits, otherwise escalate. Over time you expand coverage from high-risk lanes outward.

Expand full comment
Mark Palmer's avatar

Yes, good thoughts, although I do have line of sight into dozens of established companies and startups that are part of a "Vast Ocean" of entrepreneurs aiming to address many aspects of this field, including Snowflake who just make a major strategic investment in a company I'm on the board of (Atscale) for semantic layers, which is different than what you're talking about, but heavily related. And there are dozens of others I'm tracking doing interesting things.

It will be fun to see what emerges as the winning approach(es)! Glad to have found your Substack, looking forward to learning more.

Expand full comment