JM

NOVIS founder (@NOVISdefi) - The banking protocol for AI agents. Ever-growing, fully backed. Autonomous agents finally become economic citizens.

[JM @TheMoonMark]

Today, agents can monitor markets 24/7, rebalance liquidity across protocols, and move funds before risk events hit.

This is already happening at real scale.

[JM @TheMoonMark]

For years, DeFi reduced intermediaries — but not complexity.

Users still had to click, monitor, rebalance, and manage risk themselves.

AI agents change that dynamic.

[JM @TheMoonMark]

What’s happening in crypto + AI isn’t about “smarter bots”.

It’s about a new financial operating layer quietly taking shape.

[JM @TheMoonMark]

AI agents in crypto are no longer experiments.

They already trade, manage liquidity, route payments, and move capital with minimal human input.

That shift matters more than most people realize.

[JM @TheMoonMark]

We’re still early — but no longer guessing.
The direction is becoming clear.

[JM @TheMoonMark]

That’s the future we’re building toward at NOVIS.

I’ll keep sharing what we learn as this space evolves.

[JM @TheMoonMark]

Once money moves at software speed, a lot of old assumptions break.

Most systems aren’t ready for that yet.

[JM @TheMoonMark]

Just like APIs replaced paperwork,
and the internet removed distance,
agent-driven finance removes coordination costs from money itself.

[JM @TheMoonMark]

AI agents won’t replace humans in finance.

They’ll remove friction.

[JM @TheMoonMark]

We Built Crypto Infrastructure for the Wrong Species

I've been building in DeFAI for months now. And I've had a realization that's been nagging at me:

Multi-chain complexity is a feature, not a bug — if you're an AI agent.

We've spent years building increasingly complex infrastructure — hundreds of chains, thousands of protocols, bridges that feel like performing surgery while juggling — and we keep asking the same question: Why isn't adoption growing faster?

The standard answer: We need to simplify. Abstract away complexity.

I'm starting to think that answer is wrong. What if the infrastructure we've been building was never meant for humans at all?

The Human Breaking Point

Let me walk you through what it actually takes to optimize yield across five chains in 2026.

Let's say you have $10,000 in stablecoins and you want to earn the best yield.

Here's what you actually have to do:

Check yields on Ethereum mainnet (Aave, Compound, Morpho)

Check yields on Arbitrum (GMX, Camelot, Radiant)

Check yields on Optimism (Velodrome, Synthetix)

Check yields on Base (Aerodrome, Moonwell)

Check yields on Solana (Drift, Kamino, marginfi)

Found the best rate? Great. Now:

Acquire gas tokens for each chain (ETH, ARB, OP, SOL)

Find a bridge (which one? Stargate? Across? LayerZero? Each has different fees, speeds, and risk profiles)

Wait for confirmations

Approve tokens on the destination chain

Deposit into the protocol

Set up monitoring (because yields change constantly)

Repeat when conditions change

And this is the simple case. We haven't even touched leveraged strategies, LP positions, or anything involving derivatives.

A typical cross-chain DeFi interaction involves acquiring the destination chain's gas token, finding a reliable bridge, waiting for confirmations, and then interacting with the target protocol. Friction everywhere.

An active DeFi investor today may need to monitor dozens of protocols across multiple blockchain networks, constantly compare changing yields or interest rates, and manually execute numerous transactions. Even for experienced users, DeFi investing is mentally taxing, time-consuming, and fragmented.

The result? Most people don't bother. They leave money on Coinbase earning 4% and call it a day.
The Fragmentation Tax

The numbers tell the story.

When the same asset exists across multiple chains — like USDC on Ethereum, Polygon, and Solana — each instance creates separate liquidity pools that cannot efficiently interact. This fragmentation leads to price discrepancies, higher slippage for traders, and reduced capital efficiency across the entire ecosystem.

You're paying a tax on every trade because liquidity is scattered across 200+ chains.

For everyday users, liquidity fragmentation translates into a poor user experience. Finding the best price for a trade requires checking multiple platforms, understanding different protocols, and often bridging assets between different chains.

Complexity management is perhaps the greatest challenge facing cross-chain applications.

We've created a system where:

The best opportunities exist across chains

Accessing them requires expert-level knowledge

The friction means most value goes uncaptured

Arbitrage exists that no one can efficiently capture

Unless you're not human.

We built the perfect playground for autonomous agents and then wondered why humans found it confusing.

The Reframe

The old narrative: "We need to simplify crypto for humans."

The new narrative: "We need to build better interfaces between humans and agents who can handle the complexity."

DeFi's complexity — from wallet management to cross-chain navigation — has capped adoption to a fraction of centralized exchange volumes. AI agents can bridge this gap with intuitive, conversational interfaces that automate complex on-chain interactions, akin to having a private banker at a click.

For end-users, AI agents can abstract away today's complexity — turning wallet management, swaps, and cross-chain bridging into simple, conversational experiences. This could finally make DeFi accessible to a mainstream audience.

The multi-chain fragmentation isn't a bug to fix. It's an opportunity surface that only agents can efficiently capture.

The Agent Economy
Here's where I think this is going.

Right now, most people interact with DeFi directly (poorly) or not at all.

In 12-24 months:

Most retail users will interact through agents

Agent-to-agent transactions will exceed human-initiated ones

Cross-chain arbitrage will be fully agent-captured
Yield optimization will be automated by default

Protocol liquidity will flow to agent-friendly designs

For institutions, autonomous agents enable continuous, data-driven portfolio management, liquidity optimization, and risk controls that operate at machine speed — capabilities impossible to match with human teams alone.

The infrastructure we built for "permissionless finance" will become permissionless for agents. Humans will be the principals. Agents will be the operators.

tl;dr: Multi-chain crypto infrastructure is too complex for humans but perfect for AI agents. The fragmentation we've been trying to "fix" is actually an opportunity surface that agents can capture. The future isn't simplifying crypto — it's building the agent layer that lets machines navigate complexity while humans specify outcomes.

[JM @TheMoonMark]

This shift won’t feel dramatic at first.

It will arrive gradually — and then suddenly feel obvious in hindsight.

[JM @TheMoonMark]

These problems take time.

They don’t trend well on X.

But they’re where durable value usually gets built.

[JM @TheMoonMark]

The hardest problems aren’t glamorous:

Trust between agents.

Settlement at scale.

Identity and reputation for non-human actors.
Compliance without killing speed.

[JM @TheMoonMark]

Most AI–crypto projects today live at the application layer.

That’s fine.

But applications are replaceable. Infrastructure compounds.

[JM @TheMoonMark]

The goal is simple:

Let agents handle payments, treasury, and settlement automatically — while humans stay in control where it actually matters.

[JM @TheMoonMark]

We’re building banking infrastructure for AI-native finance.

Stable. Programmable. Boring in the right ways.

[JM @TheMoonMark]

What looks like a niche developer feature today often turns out to be tomorrow’s default infrastructure.

Agent-native payments feel like one of those moments.

[JM @TheMoonMark]

At NOVIS, we’re not building another trading product.

And we’re not building another wallet.

[JM @TheMoonMark]

When AI systems can pay other AI systems directly, without human intervention, finance stops being human-paced.

That’s a bigger shift than most people think.

[JM @TheMoonMark]

This is why agent-to-agent payments matter.
Not as a demo.

As infrastructure.