"A human is accountable for AI mistakes."
I hear this on every panel. It isn’t a usable answer.
Which human?
An agent gives a customer a €2,000 discount nobody intended...
and the finger-pointing starts.
The vendor points at the bank's spec.
IT points at the vendor.
The business points at the model.
Nobody asked Risk.
When "a human" is accountable, nobody is.
So half the room jumps to the other extreme: "Just don’t use AI."
Wrong conclusion. Companies already solved this for people.
Give the agent what a new clerk gets on day one:
- a hard limit, per decision and per day
- a named person who signs that limit
- a human who decides everything above it
- a receipt for every action
The question was never "AI or no AI".
It’s which decisions an agent can take alone, up to what amount and on whose signature.
Then "a human" has a name and a number.
Live on NEAR testnet and mainnet. Try it:
Check a stamp: https://t.co/Lseh1MTPCs
Android app: https://t.co/8Hdclxtqui
Code: https://t.co/CQL3DjGQOb
Would you trust a stamp like this on a reply?
Anyone can paste AI text and call it theirs. I typed this by hand on NTK, a keyboard that seals how I type in an enclave on NEAR and stamps my text. [NTK eae09a1a16]
Built at #NEARHackathon@NEARProtocol@NearBuilders
Privacy: your typing fingerprint stays sealed in the enclave, where nobody can read it, not even you. A stamp holds a hash of the text, never the text or your timing.
They *almost* nailed it this time around. Not sure why they insist adding "fake thickness" to their designs. At least now its only around the back vent. It looks awful. Whats the word opposite to 'elegant'?
@McLarenAuto#Artura
Stategram is the contract between human and AI.
Public beta available now for fintech and insurtech teams who want to get their agents past risk review.
A demo of how to set up a daily limit across all agents:
Pro tip for founders:
Add an appendix slide to your pitchdeck with all of the relevant information about the company in tiny letters. Leave the rest of the deck short and impactful.
Most investors are using LLMs to read it. The ones that don't benefit from minimalism.
@Tesla why doesnt it have thrusters pointing up to avoid landing on the roof. Would be a pity to lose your precious toy, no?
ALso it would make it more badass.
I was very surprised to learn that 9 in 10 executives say AI has not changed their firm's productivity in 3 years.
AI can do the work. The decisions about money need an owner, and it is never the AI.
The bottleneck was never capability. It was accountability.