Zi Xuan 子萱

I've lost count of how many crypto projects have slapped "AI" onto their roadmap over the past year. Apparently every protocol now has an agent that will trade better than you, manage your portfolio, and somehow never make a bad decision. Maybe some of them will. Most won't.
The bigger issue isn't whether AI can click buttons faster than humans. That's already solved. The uncomfortable part is handing an algorithm permission to move your money and hoping it behaves.
Newton Protocol is trying to deal with that instead of pretending the trust problem doesn't exist.
The basic idea is simple enough. You don't give an AI unlimited control over your wallet. You set boundaries first. The agent operates inside those limits, and the protocol is supposed to produce cryptographic proof that it actually followed the rules. That's where things like Trusted Execution Environments and Zero-Knowledge Proofs come in. They're not there because they sound impressive. They're there because "trust us" has aged badly in crypto.
Newton is also building its own rollup around this idea. Not another general-purpose chain trying to do everything, but infrastructure aimed at AI-driven automation. On top of that sits a marketplace where developers can publish AI agents for trading, portfolio management, yield strategies, and other repetitive on-chain jobs.
Whether that marketplace ends up full of genuinely useful tools... that's another story.
Crypto has a habit of building marketplaces before there's anything worth buying.
The NEWT token isn't particularly surprising. It handles staking, governance, validator incentives, fees, and network security. Pretty standard. I care less about the token design than whether people actually use the network. A token can have perfectly balanced economics on paper and still spend years drifting sideways because nobody needs it.
The project naturally gets compared with names like Autonolas, Fetch.ai, and Safe. I don't think Newton is really competing on who builds the smartest AI. Its bet is different. It assumes that as AI starts handling larger amounts of capital, proving what the software actually did becomes more valuable than making bigger promises about what it could do.
That feels like a more realistic direction.
There are encouraging signs. Major exchange listings helped put it on the map, and contributors connected to Magic Labs add some credibility. Still, exchange listings are marketing events, not proof of demand. Crypto confuses those two things all the time.
I also don't think the hard part is the cryptography.
The hard part is convincing developers to build here instead of somewhere else. Good infrastructure doesn't automatically attract users. We've watched technically solid projects disappear because they never reached that critical point where people kept showing up without incentives.
So I'm cautiously interested.
Not because Newton has AI in its name. Honestly, that's the least interesting part. The more interesting question is whether crypto finally starts treating autonomous software the same way it treats smart contracts—something that should be verified, restricted, and predictable instead of blindly trusted.
If that shift happens, Newton could end up being more relevant than people expect today.
If it doesn't, it'll probably become another technically impressive project that solved a real problem before the market decided it actually cared.
@NewtonProtocol $NEWT #Newt

There’s this assumption floating around right now — mostly pushed by AI grifters and VC decks — that the future of crypto is autonomous.
Your wallet trades for you.
Your AI farms yield.
Your AI rotates your stables.
Your AI protects downside.
Yeah, maybe.
But every time I hear that, I keep thinking the same thing:
Who exactly is insane enough to hand over capital to an agent without guardrails?
That’s the part people skip over.
Because crypto loves skipping the ugly part of a story.
We all talk about AI like it’s some inevitable upgrade. Nobody talks enough about how fragile that gets the moment real money touches it. Not testnet money. Real size.
And Newton Protocol... I’ll give them this — they seem to understand that.
That alone makes them more interesting than most of the AI token pile.
Not because they built some genius model. They didn’t.
Because they’re focused on constraining the model.
Big difference.
Most people won’t catch that.
Newton, if you strip away the branding, is basically trying to build a permission layer for autonomous on-chain actions.
That sounds dry. It is dry.
But it matters.
The idea is simple enough: you tell an AI what it’s allowed to do, set boundaries, and it operates inside them.
Not “do whatever.”
More like:
you can move 20% of my stables into lending markets,
only these protocols,
max 3% slippage,
don’t touch illiquid garbage.
Stuff like that.
That’s actually practical.
Because the truth is, most AI in crypto right now is either glorified dashboards or sentiment bots pretending they’re useful.
Newton is aiming at execution.
That’s harder.
Way harder.
And riskier.
People compare this to automation protocols, but I think that comparison misses the point.
Chainlink Automation exists.
Gelato Network exists.
Even old stuff like Keep3r did parts of this years ago.
But those systems are basically condition-checkers.
If X happens, do Y.
That’s not intelligence.
That’s scripting.
Newton wants something looser. More adaptive.
Like:
find best yield opportunities across chains, assess risk, rebalance over time.
That’s not a script anymore.
That’s closer to delegated judgment.
And delegated judgment in crypto is... ugly.
Because judgment can be bad.
Very bad.
People forget this.
This is where Newton’s design gets more serious than the usual AI coin nonsense.
They combine secure execution environments with zero-knowledge proofs.
On paper, it sounds strong.
In practice? Depends.
TEE keeps the computation isolated.
ZK proofs let you verify outcomes.
Fine.
But here’s where I get skeptical.
Verification sounds nice until you realize what’s being verified.
Execution.
Not quality.
That’s the hole.
A strategy can follow every rule perfectly and still bleed money for six months straight.
That’s not a bug.
That’s markets.
So anyone pretending “verifiable AI trading” solves trust completely is overselling it.
Newton reduces one category of risk.
It does not remove bad decisions.
Important distinction.
A lot of people are going to learn that the hard way.
The marketplace angle is interesting though.
Messy, but interesting.
Developers can launch agents. Users pick them. Fees get shared.
Basically strategy-as-a-service.
I can see why that catches attention.
But man... crypto has a terrible history with open marketplaces.
Absolute garbage floods in fast.
Always.
You’ve seen it before.
Yield vaults in 2021.
Signal groups in 2022.
Telegram bots in 2023.
Same cycle.
Ten decent products buried under hundreds of useless ones.
Newton’s marketplace could end up looking exactly like that if they don’t get strict.
And strict means hurting feelings.
Bad agents need to get slashed. Hidden. Punished.
Otherwise it becomes noise.
More noise is the last thing this space needs.
Then there’s NEWT itself.
Token does what tokens usually do.
Pay fees.
Stake.
Governance.
Collateral.
Nothing weird.
Supply sits at 1 billion.
Circulating is low enough that price can swing hard if narrative catches.
And narrative definitely matters here because AI is still hot.
Sometimes irrationally hot.
Honestly, a lot of these AI coins are trading like people forgot how little revenue exists underneath them.
Newton could get pulled up by that.
Or crushed by it.
Depends how the cycle behaves.
That’s crypto.
Half fundamentals, half mood swings.
Maybe more mood swings.
The thing I keep asking is whether this is too early.
Not wrong.
Early.
Those are different.
Because the demand for autonomous finance is still mostly theoretical.
Retail isn’t asking for this yet.
Most users can barely manage a bridge without stress.
Now we want them configuring agent permissions and risk parameters?
Come on.
That’s not mass market.
Not yet.
But treasury managers?
Funds?
DAOs?
That’s where this gets real.
That’s where Newton could actually fit.
Smaller user count. Bigger capital.
Makes more sense.
And that’s probably the whole bet.
Not “AI for everyone.”
More like AI for serious capital first.
That’s smarter.
Less flashy too.
If they execute, Newton could become one of those invisible pieces of infrastructure that ends up everywhere.
Nobody talks about it much, but it’s underneath everything.
That’s the best-case.
Worst case?
Chainlink notices the opportunity, builds its own version, and Newton gets squeezed into irrelevance.
That’s not hypothetical.
That’s how crypto infrastructure dies.
Quietly.
Good ideas don’t guarantee survival here.
Never did.
Still don’t.
But at least Newton feels like it’s solving an actual problem instead of stapling “AI” onto another token and calling it a revolution.
That bar is low.
Painfully low.
And yet here we are.
@NewtonProtocol $NEWT #Newt