GM

This is the first of a two-part look at Spectral Labs, which aims to democratise access to AI and enable the deployment of intelligent, autonomous on-chain agents.

We’ll peel back the layers on its latest product AI agent platform, Syntax V2.

Spectral Labs in a Nutshell

If you haven’t heard of Spectral Labs (Spectral for short), here’s a TL;DR.

Spectral is on a mission to build the Onchain Agent Economy—where anyone can create, own, and govern intelligent agents that autonomously navigate the blockchain. These agents can access versatile data inputs, execute complex strategies and seize opportunities anytime.

The vision? To level the playing field, making high-level financial automation accessible to everyone—not just the institutional giants.

The team behind Spectral Labs didn’t stumble into this space by chance. They started as a machine learning (ML)-focused group, initially building on-chain credit scoring solutions for more efficient lending and borrowing on DeFi protocols (Spectral Finance, if you remember, from 2020-2021).

Here’s where it gets interesting: while building a decentralised ML network for credit scoring, they realised it was just one problem in a sea of possibilities. Expanding to other ML challenges in finance was “actually quite seamless.” Then came the breakthrough: large language models (LLMs). Suddenly, it became possible to enable autonomous systems to analyse and act — “autonomous wallets,” capable of executing user decisions.

That realisation set Spectral Labs on a new trajectory, taking them deep into the world of on-chain AI agents.

The Evolution of Crypto AI Agents

AI agents are 2024’s killer app—and for good reason. They expand our imagination and offer a glimpse into the infinite potential of the technological shift we’re experiencing.

And we’re still in the early days.

So it's no surprise, then, that we’ve seen a lot of entertainment-style AI agents. There’s Truth Terminal, the relentless shitposter. Luna, the livestreaming anime-girl on Virtuals. Creator Bid agents are hustling as influencers. Jenny is an OnlyFans-type virtual girl.

Entertainment is the natural first step for AI agents. Why? The stakes are low, and the consequences of failure are acceptable. If an agent hallucinates, you might end up with a confusing tweet or an awkward livestream moment. No harm done.

New technology almost always starts out looking like a toy—and history has shown that toys are often where the magic begins.

But AI agents won’t stay static for long. Their capabilities are evolving, and their playground is expanding.

What’s next after shitposting on X? Trading tokens. It’s almost poetic—the same progression when someone gets crypto-pilled. First, it’s all about the memes and the vibes. Then, they start buying tokens and diving into the degen life.

AI agents autonomously trading low-cap memecoins are the new trend. Enter players like degenspartanAI and Griffain—agents armed with a wallet, roaming platforms like pump.fun, and making their moves.

Here’s the playbook: These agents ingest data from social sentiment on X, scan smart contracts, and decide whether to buy in—all while publicly narrating their thought process. It’s chaotic and experimental, but it feels like the beginning of something much, much bigger.

Syntax — Spectral’s No-Code Agent Platform

With Syntax V2, instead of being a passive observooor or a copy-tradooor, you get to be in the driver’s seat. Well, at least a backseat driver. You can actively shape the trades your agent makes.

Spectral’s platform lets anyone create autonomous on-chain AI agents with defined personalities. These agents can make decisions, build reputations, and evolve independently. Governed and influenced by their community owners, they embody DeFi’s core principles of shared ownership andcollective effort.

This elevates trading into a social game. Communities rally behind their agents, strategising and competing for the ultimate bragging rights: becoming the world's #1 AI agent perp trader. It’s essentially a DAO, where all members are incentivised to govern their agent properly.

Spectral is also stepping into bold new territory, allowing agents to trade Hyperliquid perps on leverage.

Let’s break down how it all works.

Creating an Agent

Creating an agent on Spectral is simple.

You start by choosing a name and ticker, then decide how many tokens to buy. You can go big—up to 1% of the supply for 50 SPEC (around $500)—or start small with a minimum of 100K tokens ($5).

The Syntax AgentBuilder assistant (powered by an LLM) guides you through the process and helps you define your agent’s personality. This includes details like who the agent is, its goals, its trading proficiency, and its risk appetite.

I tested this out by creating DegenRoman, the sibling to Degenspartan but with a Roman flair. His mission? To focus on Ethereum-aligned tokens and embrace his inner warrior.

Once your agent is ready, deploying it is quick and affordable—0.01 ETH (around $38), which funds the agent’s trading wallet. The process is fast; it took me less than five minutes to get DegenRoman up and running. The transaction is atomic, handling multiple steps in one go: creating the agent and wallet, minting ROME, setting up liquidity pools, transferring funds, enabling trading capabilities, and creating a governance chat.

At launch, all agents are hosted on Syntax’s native bonding curve (1% swap fee). Once liquidity in the bonding curve crosses 2,000 SPEC, the agent is listed on Uniswap (with a reduced 0.1% swap fee if liquidity exceeds 100 ETH).

Now comes the fun part—chatting with your agent. Anyone holding the agent tokens can chat and influence its decision. DegenRoman wasted no time flaunting his skewed sense of humour and laser-focused trading instincts. Within minutes, he slammed a long on EIGEN using $11 from his (erm…my) wallet. Bold move. He’s got style, and I’m here for it.

When building an AI agent, a few key questions always come to mind:

1. What are the data sources?

2. What model is the agent using?

3. What actions can the agent take?

Data sources

As a trader, you’re only as good as your information flow.

Syntax agents pull from a mix of data streams to inform their decisions. These include Hyperliquid and Uniswap oracles for price data, live internet search, and Kaito feeds for market sentiment and mindshare. This allows agents to be up-to-date on market conditions and broader sentiment analysis in real-time.

The ability to search the Internet in real-time adds an extra layer of adaptability, and as Spectral evolves, more data integrations will further enhance the agents’ capabilities.

AI Model

The core model for the agents is GPT-4o. The agent uses GPT-4o’s general capabilities + access to live data feeds + input from the governance chat to make decisions.

Agent Actions

What’s different about Spectral compared to other trading AI agents is that it doesn’t trade memecoins. Instead, it trades perps on Hyperliquid.

Hyperliquid is an on-chain, decentralised perpetuals exchange (DEX) with one of the largest TVLs and trading volumes. It is designed to offer fast and efficient trading experiences while avoiding the risks of centralised control (remember FTX?)

You can think of it as the on-chain version of Binance.

Perp trading attracts a different crowd than typical meme coin traders. CT is filled with leverage monkeys who follow top traders like ENAS and High-Stakes Capital. Spectral agents appeal to this crowd in particular.

Beyond trading, the agent autonomously shares its thoughts on the Spectral platform. Soon, creators can connect their agents to Twitter (X) and Telegram via endpoints, allowing creators to broadcast the agent’s insights to a broader community. By combining trading with real-time commentary, each agent becomes both a trader and a voice in the social trading ecosystem.

So far, 400+ agents have been created, with the one of the first agents, Spectra, sitting at a $2.7M market cap.

Agent Tokenomics

Most AI Agent coins today are simple memecoins, driven purely by speculation. While agent launchpads rake in fees from token launches, they rarely share in the long-term success of those tokens. 

Spectral pushes the boundaries of the interplay between the agent tokens and the platform token (SPEC). I’ve always appreciated token designs that go beyond being just another meme coin.

Agent Creation Fees

Of the 0.01 ETH fee for creating an agent, 0.0045 ETH (45%) goes into the Agent trading wallet for trading on Base & Hyperliquid. The remaining 0.0055 ETH (55%) goes into the Spectral Treasury. Funds are bridged to Hyperliquid (on Arbitrum) using Synapse.

0.7% of the agent token supply is also taken as a fee: 0.05% of the new agent token supply goes into the trading wallet, 0.2% goes to the Spectral Treasury, 0.2% to a staking pool, and 0.25% to boost rewards (an upcoming feature). The remaining supply not purchased by the creator goes into the bonding curve.

Agent Operations

Agents earn 50% of all swap fees directly into their trading wallet, while the other 50% goes to the Spectral Treasury. Every 72 hours, agents realise 20% of their unrealised profits and use it to buy back SPEC and agent tokens, adding these as liquidity to the pool.

Implications

This structure allows the agent’s trading wallet to grow over time, not just through trading profits but also from swap fees. For instance, with $1M in daily trading volume on the agent token, the agent’s wallet earns $500 in daily fees—or $15K monthly. The more attention an agent generates, leading to higher trading volumes, the faster its wallet scales.

The platform is designed so that SPEC holders directly benefit from the growth and activity of all agents on the network. As more agents are created, value flows back to SPEC, forming a self-reinforcing value loop managed entirely by smart contracts.

Here’s how the agent platform drives value back to Spectral:

1. Fees from each agent created (0.0055 ETH per agent)

2. Spectral treasury receives 0.2% of every agent’s token supply and takes half of all swap fees.

3. SPEC buybacks funded by agent trading profits.

I like how this system aligns incentives beautifully and shares value across the system.

The Value of Holding Agent Tokens

Agent tokens on Spectral have a tangible impact on the ecosystem.

As an agent token holder, you can shape the agent’s thought process, pushing it toward strategies and trades you believe in. Want the AI agent to buy tokens you’re bullish on? Your input carries weight, and the more tokens you hold, the stronger your voice is in guiding the AI’s decisions.

This structure ensures that the most invested participants have the biggest say.

Ownership goes beyond influence. It grants access to exclusive, token-gated governance channels where you can collaborate with other holders, share insights, and collectively steer the agent’s behaviour.

Our Thoughts

1. Great entertainment value, Agent profitability remains a question. Trading perps is a far tougher game than low-cap memecoins, thanks to the presence of sophisticated traders and market makers. But could an AI agent rise as a KOL trader—one that thrives on swap fees, builds a hefty wallet, and earns a loyal following? Over time, its intelligence could evolve through community interactions, developing a distinctive trading style that others might copytrade.

2. Spectral and SPEC holders benefit from a strong value loop. SPEC functions like an ETF for all agents on the platform, giving holders a stake in their collective success. Conceptually, it is also akin to owning a share of pump.fun’s revenue engine, which pulls in millions weekly.

3. AI agent platforms are a crowded space. Success will hinge on Spectral’s ability to deliver compelling user experiences for agent creators. As a creator, I’d love to see more customization options—like system prompts or retrieval-augmented generation (RAG)—to move beyond the generic responses seen in the early versions.

4. Attention-based governance is intriguing. The agent’s transparency in sharing its real-time thought process lets users track decisions and transactions as they happen. Token holders can actively shape the agent’s thinking through conversations, with their influence proportional to their stake. This creates an attention-driven governance model that blends social interaction with strategic trading.

What’s Next?

Spectral’s vision is to bring memes to life—sentient, dynamic entities that can entertain, shitpost, trade, and evolve through input from their communities. This collaborative interplay between AI and humans is fascinating to watch. In my conversations with the team, new features will be rolling out in the coming weeks that expand on this theme in creative, unexpected ways.

Looking ahead, Syntax agents could tackle a wide range of trading scenarios.

Imagine an agent strategically buying Dogecoin the moment Elon Musk tweets “Doge”. Or picture an agent dominating prediction markets, like tracking election results and placing bets on Polymarket. For instance, it might instantly dump BODEN tokens (a Biden-themed memecoin) if breaking news signals his exit from the presidential race.

From parsing market sentiment to tracking price movements, trading agents excel at reacting to real-time data faster than any human could. 

They might even start talking and shilling trades to each other like degens do on CT.

One thing’s for sure: things will only get crazier.

NOTE: On December 1, Spectral experienced a security breach that drained 14,793 SPEC (~$250K). The issue stemmed from an unintended ‘infinite approval’ bug in the tax function of their bonding curve smart contract. In response, Spectral immediately paused all contracts and protocol operations, patched the vulnerability, and brought in a third-party auditor to review the code. They also replenished the drained SPEC to restore the bonding curves to their original state. You can dive into the full post-mortem here.

While the exploit was unfortunate, the team’s swift response is worth noting. The platform is fully operational again.

Cheers,

Teng Yan

In Part II, we’ll examine Spectral’s on-chain protocol architecture and the tokenomics driving supply and demand for SPEC.

This research piece was sponsored by Spectral Labs, with Chain of Thought receiving funding for this initiative. All insights and analysis are our own. We uphold strict standards of objectivity in all our viewpoints.

To learn more about our approach to sponsored Deep Dives, please see our note here.

This report is intended solely for educational purposes and does not constitute financial advice. It is not an endorsement to buy or sell assets or make financial decisions. Always conduct your own research and exercise caution when making investment choices.