Can EigenLayer carry AI agents on its network? Founder Sreeram Kannan thinks so

EigenLayer may bring its data availability service to AI agents, giving them on-chain abilities. The founder of EigenLayer believes AI agents can become part of the Autonomous Verifiable Services (AVS) ecosystem. 

EigenLayer can potentially join the AI narrative trend as a pivot strategy after attention shifted away from Data Availability technologies. The founder of EigenLayer, Sreeram Kannan, explained how AI agents can start building real on-chain activities, by becoming a part of the Autonomous Verifiable Services (AVS) network. 

Kannan believes AI agents are the ultimate AVS consumers and should join up as verification entities themselves. He explained that agents are not yet allowed to run operations in a permissionless, verifiable system. According to him, agents can become truly sovereign without suffering from potential flawed oracle attacks with EigenLayer. 

Agents are the ultimate AVS consumers
———————-
Agents should run as Autonomous Verifiable Services (AVS).
———————-
Why?
To get unstoppability, you need the agent to be able to run computation in a permissionless manner.
Since the “agent” is delegating… https://t.co/3ce3OmFqmc

— Sreeram Kannan (@sreeramkannan) January 5, 2025

Kannan has previously predicted the role of EigenLayer for AI agents, as they are too computationally heavy to run fully on a blockchain. However, their actions can be verified in another layer, creating a toolset to perform truly autonomous actions. 

EigenLayer may also be moving to reposition itself with a new narrative after the Data Availability layer has faded in terms of traction.

Since AI agent platforms are some of the most widely traded and trending tokens, some of the EigenLayer developers started testing a new marketplace. If the idea pans out, EigenLayer may compete with Virtuals Protocol, Eliza, and other platforms for hosting a new wave of individual agents. 

Following the news that AI agents could be coming to EigenLayer, EIGEN tokens traded near a one-week high of $4.12.

Currently, AI agents are mostly used for social media content, but they only have hypothetical tools to make on-chain changes. Avoiding errors and unwanted effects may be achieved through AVS calls, which have built-in protection against flawed or malicious actions. 

AI agents still cannot perform independent actions

The big problem with the wave of new AI agents is that they are not truly agentic on-chain, and rely on several allowed actions, mostly based on language models. While successful as meme tokens, AI agents are still trying to build a toolset for safe, verifiable actions. 

AVS go beyond nodes or ETH stakers and can bring a variety of on-chain actions. Those actions include the creation of sidechains, new virtual machines, oracles, bridges, privacy protocols, and cross-chain transfers. 

An AI agent can be linked to a service, which would be secured through EigenLayer’s already staked ETH. The ETH stake is valuable enough to ensure all actors coordinate since malicious actions would lead to stake slashing. 

At this point, it is unknown if any AI agent can perform on-chain actions or offer a service. However, with EigenLayer, the barrier to entry may be lower as AVSs were specifically created to allow easier access for developers. 

The actual actions of agents are still being tested, as one user added browser-based activities and even interactions with EigenLayer. There are also attempts to engineer new AI agents with sufficiently verified data and computation, to create an autonomous actor that can manage portfolios.

Once again, EigenLayer plays a critical role in ensuring that every step of AI decision-making is solid, correct, and unattackable. 

AI agents rally to higher valuation

After turning into one of the most successful narratives for 2024, AI agents continued their climb. 

All AI tokens expanded their valuation to above $16.9B, as traders attempted to discover the next best agent. The competition was also for AI agent platforms, each taking a different approach to building new models. 

Trending tokens included Freysa (FREYSA), an interactive AI personality with puzzles, and Swarms (SWARMS), a new mechanism to create AI agents. 

There is no set standard on what the agents can do, and the sector is still in its experimental stage. EigenLayer has no official partnerships and has not endorsed any specific agents. EigenLayer has not officially announced the launch of an AI agent framework, but it has hinted at building its own AI agent universe. 

Launching an AVS agent would raise demand for staked ETH. AVS entities are secured by more than $16B in staked ETH, which means even tokenless agents can be more secure. 

From Zero to Web3 Pro: Your 90-Day Career Launch Plan