@reidhoffman The Anthropic/FIS bet is real. But the trust layer that makes it defensible in a court, in a compliance review, or in front of a regulator is still wide open.
@reidhoffman That infrastructure — a live, independent truth layer that continuously watches what agents believe, detects contention, and resolves it through declared organizational authority — is closer to an independent auditor than a logger. And it doesn’t exist yet.
@reidhoffman It’s a deeper question: were the facts the agent consumed when it made that call verified, current, and authoritative? A perfectly accurate model reasoning from stale or contested facts is still a liability. No one is governing the inputs — only scoring the outputs.
@reidhoffman They’ll build a layer that is structurally independent — one agents cannot manipulate or bypass, that is never an add-on.
Because the AML question isn’t really 64% vs 85% accuracy.
@reidhoffman The governance layer can’t “sit around” the agent. If the agent can route around it, it will. That’s not a product gap — it’s a category error.
The startups that win in regulated AI won’t build explainability tooling or audit logs bolted onto the side.
A rocket that Hamas terrorists launched towards innocent Israeli cities fell short and exploded on hospital in Gaza city causing extensive destruction and hundreds of deaths.
The Hamas are killing the citizens of Gaza !
Share!!!
#hamasiISIS
היה פה 6 שבועות להכין משהו כזה. היה זמן לעשות פיילוט בכמה יישובים ולראות שיעורי הדבקה. היה זמן להכין את הגנים ובתי הספר בצורה מסודרת. כל יום של הסגר עלה מיליארד וחצי ש"ח אומרים. מה הם עשו? כלום וחצי. כמו שרצו לסגור, רצים עכשיו לפתוח.
Expanding to a new audience can be an effective way to grow your business—but make sure your next target audience is worth the investment https://t.co/sR0IHZnr6O
Brands constantly try to guess what's next in marketing. It makes the most sense to meet consumers where they are -- on their phones watching videos. https://t.co/QueoIqcJkw