ZIV OREN

I didn't start looking into Newton Protocol because I was searching for another AI project. What actually pulled me in was a small question that stayed with me after reading through its design. I kept thinking about where the real decision-making ends and where the infrastructure begins. At first, I assumed that distinction would be obvious. It wasn't.
The more I looked, the more I realized I had been treating AI as the main story. That assumption didn't last very long.
Like many people following Web3, I've seen plenty of projects promise AI-powered automation. Most discussions eventually revolve around how intelligent the models are or how accurately they can make decisions. My first thought was that Newton Protocol was competing in exactly the same space. I expected the focus to be on building smarter agents for trading, strategy execution, or automated financial management.
As I spent more time reading, that idea slowly changed.
What stood out to me wasn't the AI itself but the environment the AI is expected to operate in. Newton describes itself as building a secure rollup for AI-driven strategies, automated trading, and a marketplace where developers can publish and monetize those strategies. Initially, I treated those as separate features. Eventually, I started wondering whether they're actually different pieces of the same idea.
If autonomous systems are going to control financial activity, then intelligence alone probably isn't enough. Someone has to trust how those systems operate, not just what they decide.
That seems like a much harder problem.
I found myself imagining a simple scenario where an AI strategy decides to execute a trade. From the outside, it might look like one decision followed by one transaction. But the more I thought about it, the less believable that became. Between the AI generating an action and that action appearing on-chain, there are several layers involved. There are execution rules, verification, rollup infrastructure, settlement, and whatever policies determine what the AI is actually allowed to do.
From a user's perspective, all of those layers disappear into a single outcome.
That made me stop blaming—or crediting—the AI for everything. If an automated strategy reacts differently than expected, I can't say for certain the explanation starts with the model. It could just as easily be the infrastructure enforcing security checks or operational constraints before anything reaches the blockchain. Different components could easily produce the same visible result.
I think that's an important distinction because discussions around AI often skip over everything happening behind the scenes.
Another thing I kept coming back to was incentives.
Crypto generally rewards speed, openness, and permissionless participation. AI, on the other hand, rewards adaptability and continuous learning. Financial infrastructure usually values something different altogether. It prefers predictability, clear rules, and systems that behave consistently under pressure.
Those incentives don't naturally fit together.
The more I looked at Newton, the more it seemed like an attempt to balance those competing priorities rather than maximize just one of them. I don't know whether that balance will work in practice, but it feels like a more realistic challenge than simply building increasingly capable AI agents.
The marketplace for AI developers also became more interesting the longer I thought about it. My initial assumption was that it would simply be a place where developers publish strategies for others to use. That now feels like only part of the picture.
If hundreds or even thousands of AI strategies eventually exist in one ecosystem, choosing between them becomes its own challenge. Performance will matter, but I doubt it will be the only thing people care about. Reputation, transparency, consistency, and risk management may become just as important. An AI strategy that performs well during favorable conditions isn't necessarily the one people will trust with meaningful capital over time.
Of course, that's speculation on my part. I haven't seen enough real-world activity to know whether those incentives will actually emerge.
I also found myself thinking differently about the secure rollup.
At first, I viewed it as another scaling solution. That felt like the obvious interpretation. But the longer I considered the architecture, the less convinced I became that scaling is the most interesting part. Maybe the rollup matters because it provides an execution environment where autonomous systems can operate within rules that are transparent and verifiable.
If that's the case, then the infrastructure isn't just supporting AI. It's quietly shaping how AI is allowed to participate in financial systems.
That idea seems more significant than it first appears.
One thing I still can't answer is how users will eventually assign responsibility. If an AI strategy behaves unexpectedly, where does accountability actually begin? Is it the developer who designed the strategy? The protocol defining execution rules? The rollup processing transactions? Or the marketplace distributing the strategy in the first place?
It's tempting to search for one simple explanation, but I suspect there isn't one.
The more layers a system contains, the harder it becomes to isolate the source of any particular outcome. What users experience as a single action may actually be the product of several independent mechanisms working together.
That realization changed how I think about projects like Newton Protocol. Instead of asking whether its AI is smarter than everyone else's, I started asking whether its architecture makes autonomous finance more trustworthy without removing the flexibility that makes AI useful in the first place.
I honestly don't know the answer yet.
Like most early infrastructure projects, a lot of the interesting questions won't be settled until developers begin building at scale and users begin trusting real value to automated systems. Designs that look convincing on paper sometimes struggle under real-world conditions, while overlooked architectural decisions occasionally become the features that matter most.
For now, I find the infrastructure more interesting than the intelligence.
AI models will continue improving regardless of which protocol wins. Better models will arrive, costs will change, and today's competitive advantage may disappear surprisingly quickly. Infrastructure tends to evolve more slowly, and when it works well, most people barely notice it.
Maybe that's where Newton's real experiment lies.
Not in proving that AI can make financial decisions, but in exploring whether autonomous systems can operate inside an environment that people are willing to trust over the long term.
I'm still exploring that question myself, and I don't think there's enough evidence yet to reach a confident conclusion. But it's the question I kept coming back to after spending time with the project, and I'd genuinely be interested to hear how others interpret the architecture because I suspect there are angles I haven't considered yet.
@NewtonProtocol #Newt $NEWT