A few months ago I probably would have judged an AI project with one question:
Is the model good?
Now I'm not so sure that is the right question anymore.
The more AI people talk about agents, automated trading and strategies running by themselves, the more another question keeps coming back.
How do we know what actually happened?
Not what the dashboard says.
Not what the team says.
What actually happened.
Maybe that sounds boring compared to new models and benchmarks.
Still feels important.
If an AI opens a trade and something goes wrong later, where do you even start looking?
The model? The strategy? The infrastructure? I honestly don't know.
That is probably why I stopped seeing @NewtonProtocol as another AI project.
The idea behind Newton Mainnet Beta feels closer to infrastructure.
Less about making AI smarter.
More about making AI actions easier to trust
Maybe I'm oversimplifying it.
But the longer I spend around crypto, the more I notice that markets eventually care about proof.
Proof of reserves. Proof of liquidity. Proof of ownership.
Maybe AI ends up following the same path.
Not because trust is exciting.
Because eventually trust becomes necessary.
Anyway, that was my takeaway after reading more about Newton this week.
Curious if other people see it differently.