Something about the connection between AI agents and blockchain data made me look a little closer at @SpaceandTime
An AI agent might receive a perfectly valid-looking database result and act on it without knowing whether the information was ever correct in the first place. I've been paying more attention to this gap.
Proof of SQL is an interesting part of Space and Time's approach. Rather than asking an application to accept a query result on trust, it provides a ZK proof that lets the result and query execution be verified.
I keep coming back to the AI side of this. If an agent is making decisions automatically, bad input doesn't necessarily stop at one incorrect answer. It can lead to an incorrect action.
Giving AI access to more data is one challenge. Making that data verifiable is another, and probably a less visible one.
I keep wondering whether verifiable data will become a necessity as AI agents start making more decisions without human supervision. Will AI agents ever be truly reliable without verifiable data?
Spent an evening looking through KiteScan activity and came away with more questions than answers.
Kite’s documentation describes payment channels where opening and closing happen on-chain, while thousands of smaller updates can happen off-chain.
That makes transaction counts a tricky way to judge actual agent adoption.
One thing caught my attention in the October 28–29 activity window.
A public tracker collecting KiteScan data recorded 2,133 sampled transactions involving the WKITE contract over 24 hours, compared with 608 involving SimplePair.
What stood out wasn’t the size of the numbers. It was how much visible activity involved familiar token infrastructure rather than clearly identifiable agent payments.
I can’t tell how many of those interactions came from actual autonomous agents, ordinary users, or automated scripts. The sample doesn’t establish unique wallet counts either.
Small confession: I used to judge a chain by its daily transaction count. Looking at Kite made me rethink that.
A busy network is one thing. Independent agents repeatedly paying for real services is another.
For $KITE , that’s the distinction I’d like to understand better.
The infrastructure is there, and transactions are happening. Whether those transactions reflect meaningful agent adoption is still an open question.