These days, you’d be hard-pressed to avoid the buzzword “AI.” Every Fortune 500 tech giant, crypto startup, and even your cousin’s Shopify store is scrambling to sneak it into their marketing copy.
If you don’t join the AI parade, your competitors certainly will — and when your numbers start to dip, you’ll soon realize that adding a dash of AI to your marketing might just be the ticket to keeping up.
Don’t overcomplicate it. The name of the game is improvize, adapt, and overcome.
Setting aside the satire for a moment, AI is indeed a groundbreaking technology with tremendous potential to transform our society in unimaginable ways.
While much of what's currently available may be just a passing trend, and the AI hype appears to be swelling into something that could rival the dot com bubble, we get to witness the true extent of what this technology is capable of firsthand.
However, before the onset of the AI agent craze, only a select few projects were working toward this vision.
Mode, a layer-2 (L2) network, recognized this opportunity last year and wisely chose to pivot from a general-purpose L2 to a platform that is rapidly emerging as the premier hub for AI agent innovation.
What is Mode Network? The home of DeFAI
First and foremost, Mode is an L2 network built on the Optimism stack. It features sub-cent transaction fees and high throughput, which are necessary for users and their agents to effortlessly participate in DeFAI.
AI agents, in particular, have become a hot topic within the crypto community since the release of Truth Terminal in October last year. However, it’s important to note that your average Twitter (X) bots or ChatGPT wrappers are not the topic of discussion today.
Instead, we’ll discuss AI agents capable of performing on-chain actions like investing, trading, bridging, and more on behalf of their human overlords. But what makes them different from regular bots?
A great explanation I found about what an AI agent is originates from the litepaper of an AI project that recently took part in Mode’s DeFAI accelerator Day.
‘’An Agent is a software application capable of extracting information from a dynamic and unstructured environment. It can reason about this information in relation to its objectives, discover patterns within the data, and learn how to leverage those patterns effectively. Additionally, an Agent can perform actions that its owner may not have even considered.’’
Hmm, I wonder which sector in crypto desperately needs this type of functionality.
The Mode team believes that so far, we have all been DeFi beta testers, and the ideal user is, in fact, AI.
Navigating the world of DeFi often feels like an elite club reserved for those with PhDs in computer science. To start, decentralized apps (dApps) are like hidden treasures scattered across a maze of blockchains.
Just when you think you've discovered a useful tool for spotting opportunities, you realize you also need to find the right bridge and wallet. Plus, you’ve to figure out which gas token to use for each chain.
After you've done all that, you inevitably realize that due to crypto’s volatile nature, the positions you’ve just entered require 24/7 monitoring. I could go on, but you get the gist.
None of this is a problem for AI, which is extremely good at evaluating unstructured data across many sources and executing a strategy on the user’s behalf.
DeFAI builds on DeFi's existing foundations while simplifying the user experience by making the complex intricacies invisible. New to that term? Check our idiot's guide to DeFAI for an intro.
?While true DeFAI agents are still in their early stages, thanks to open-source frameworks like ElizaOS, Arc, ZerePy, and others, developers can now integrate them into the broader DeFi landscape.
But creating an AI agent is just part of the challenge; the other part involves identifying the blockchain infrastructure that best supports it.
Infrastructure
Three distinct infrastructure layers work in tandem to enable agents to begin outtrading us, mere mortals:
- Application/agent layer
- Data layer
- Execution layer
Execution layer
After obtaining a $2 million grant in OP tokens, Mode is dedicated to bringing the DeFAI vision to fruition using Optimism’s OP Stack. This involves building on rock-solid foundations, working alongside other teams, and allocating a share of its sequencer revenue to the Optimism Collective.
Consequently, the advantages gained — such as significant liquidity and capital supported by Ethereum’s security — are undeniably valuable.
The Superchain ecosystem includes almost 30 OP Chains, including Base, Ink, Zora, World Chain, Fractal, and Soneium.
Nearly half of all L2 transactions occur on the OP Stack, making it the most widely used framework for building L2s on top of Ethereum. This alone gives Mode substantial network effects, crucial for realizing the DeFAI vision.
But, for the AI agents to operate within DeFi intelligently, they require access to trustworthy data.
Data layer
The data layer consists of two parts — oracles and a purpose-built Bittensor subnet called Synth.
This layer is crucial for enabling agents to make informed decisions. It provides complex data sources that enhance their reasoning capabilities, allowing them to forecast market conditions and adjust strategies accordingly.
Oracles can be seen as the glue that holds blockchains together. They send a constant flow of information between chains that utilize a particular oracle solution.
For maximum reliability and uptime, Mode has integrated not one but several oracle networks, including API3, Pyth, Redstone, and Supra.
Then, with the help of Synth, Mode is able to push the boundaries of what AI agents are capable of by providing them with the most accurate synthetic data, starting with Bitcoin.
You see, until now, AI agents' forecasting capabilities have been limited due to a lack of valuable data on which to train. Synth changes this by incentivizing miners to continuously submit synthetic models that capture the entire distribution of possible price movements and their associated probabilities. More on this later.
In layman's terms, AI agents will improve their ability to predict future outcomes, bringing you higher returns.
Next, we arrive at the application layer, where this infrastructure converges to deliver the future of operating on-chain.
Application layer
The application/agent layer is the real breakthrough in crypto. It offers simple interfaces for users to access various AI-operated strategies without interacting directly with complex DeFi protocols.
The best place to look for such applications is within Mode’s recently launched AI Agent App Store. With nearly $300k in Assets Under Agents, everyone can access 100+ AI-powered tools and put them to use right now.
The store features applications such as ARMA, the leading agent for maximizing yield across money markets; Olas Modius, the inaugural AI DEX liquidity farming strategist; and Brian, an agent that simplifies DeFi transactions using natural language prompts.
Given the vast array of existing AI applications and the many more on the horizon, accessing them all from a single location is a brilliant approach. However, none of this would be achievable without working together with other projects.
Ecosystem
Mode has built a robust network of AI-focused partnerships. In last year's DeFAI Accelerator, Mode chose nine innovative teams to collaborate closely with crucial contributors and partners to speed up development.
These initiatives focus on on-chain AI agent Infrastructure, training platforms for DeFi agents, AI-driven asset management, security agents, AI-generated user interfaces for protocols, verifiable AI utilizing ZK and TEE, and even no-code model training that can be accessed from your phone.
All these solutions are currently available or will soon be found in the previously mentioned AI Agent App Store.
For these projects to achieve their potential, having active DeFi protocols in place is crucial.
For Mode, this involves partnering with top industry players such as Velodrome, an advanced AMM that incorporates features from Curve, Convex, and Uniswap; Beefy, known for its automated yield farming strategies; Peapods, which offers a unique volatility farming solution; and y2k, which specializes in derivatives and hedging services, among several others.
Then, thanks to projects like Hyperlane, LayerZero, Retro Bridge, and Chainlink, users from all corners of the blockspace can access Mode’s AI agents with ease.
The full list of projects that are part of Mode’s ecosystem can be found here.
With an already burgeoning ecosystem of AI agents, DeFI protocols, and integrations from partnering AI projects, you’re probably wondering how to position yourself within this narrative.
I’m glad you asked.
Mode tokenomics
Similar to many L2 solutions, Mode includes a governance token. However, what sets it apart is its innovative approach as the first L2 to adopt a vote escrow model for distributing incentives to ecosystem protocols.
While MODE acts as a liquid exposure to Mode’s upside, veMode is where the real action can be found.
veMode
Following the introduction of Season 3 in October last year, the project adopted a vote-escrowed model for MODE, which provides several benefits compared to the traditional models utilized by most blockchains.
In this new approach, you can participate in governance by staking veMode. Users can stake veMode to influence protocol emissions with only a three-day cool-off period. Keep in mind that the longer you stake veMode, the greater your voting power.
Another significant distinction from the standard mechanism, which usually focuses incentives on liquidity pools, is that users can vote for specific protocols. This allows teams to target their growth efforts more strategically.
To recap, vote for your favorite application to direct emissions, enjoy rewards for participating, and use the application for increased rewards — all while getting the chance to qualify for an airdrop at the end of each season.
With 2 million in OP incentives, airdrops from new AI projects launching on Mode, whitelist opportunities, and access to an exclusive TG group, veMode holders are the first citizens of Mode’s DeFAI economy.
Feel free to stake your tokens, cast your votes at the voting portal, and then return to uncover the surprises that 2025 has in store for Mode!
Roadmap
Now that you're back, let’s explore the developments awaiting you in DeFAI.
While Mode has undoubtedly positioned itself as the hub for AI innovation, the work continues. The network itself is expected to receive several key upgrades that will enable even greater functionality for AI agents.
The first development we can expect this year is the Mode AI terminal, which will launch in January. From here, users will be able to use a chat co-pilot that can perform numerous DeFi actions such as swap, stake, and earn yield.
Furthermore, the terminal will also allow developers to launch smart contracts and deploy simple functions like distributing tokens to multiple addresses or creating NFTs.
One of the terminal's standout features will be its ability to address probabilistic questions using market data, a capability that no other agent or LLM can currently provide.
Synth, the specialized Bittensor subnet we referenced earlier, which is dedicated to the advanced training of agents, will enable this functionality.
As stated by mode’s Co-Founder James Ross:
‘’The vision of Synth is to start with single-asset price paths, move to multiple-asset price paths, then integrate correlated asset paths, and finally expand to generate Monte Carlo simulated data for other industries, e.g., commodities, weather, traffic, sports, lifespan, and relationship outcomes. This data enables AI agents to operate in uncertain environments and strategize with incomplete information.’’
From what I gathered, Monte Carlo implies that we will soon have enough wealth to live in Monaco.
Anyways, the subnet is currently in the testnet phase but is anticipated to launch on mainnet this year.
Another innovative feature currently in development and already available on the testnet is the first AI-secured Sequencer, created using Forta’s Firewall technology.
This system incorporates a cutting-edge AI-powered transaction risk-scoring engine. The sequencer aims to leverage simulations and AI to assess transactions prior to execution, enhancing the safety of on-chain operations.
Lastly, the imminent launch of a bridging protocol will transform the Superchain into a more unified ecosystem by resolving significant challenges such as slow bridging and fragmentation.
It will also establish standardized smart contracts for developers, enabling other L2s to better integrate with Mode’s distinctive infrastructure, and facilitating a quicker expansion of the DeFAI vision.
Concluding thoughts
Throughout 2024, we witnessed a taste of what’s to come at the intersection of AI and crypto. While Bittensor surely took center stage within this first wave of interest in AI, most of the development was still occurring on the side of infrastructure — just a stretch too far for apes like us to comprehend.
With the advent of AI agent meta, however, we have witnessed the first consumer-facing applications arrive. Now, what exactly we can expect from this intersection is beginning to crystallize, and the future is bright.
The most promising subsector is undoubtedly the one suggested by Mode. DeFAI aims to connect everyday consumers with the intricate DeFi ecosystem by offering straightforward applications that automate repetitive tasks like bridging, swapping, and scouring the web for the best opportunities.
With a portfolio of over 100 agents in the App Store and upcoming improvements to its infrastructure, Mode is poised to be a key player in the rapidly advancing AI-driven crypto ecosystem.
