Artificial intelligence steadily progresses to become an integral part of business processes. Blockchains also benefit heavily from the technology, especially AI agents that can automate all kinds of tasks executed onchain. Yet, their integration is open for improvement to lead to more efficient use cases.
Cointelegraph Accelerator recently brought together venture capitalists in an X Spaces session to explore the matter. Zoie Zhang, co-founder of Stealth Project; Fiona Ma, the investment and research lead at DWF Ventures and Samiz Bayan, an investor at Draper Dragon, discussed the intersection of AI agents and crypto, how crypto rails can enhance AI development and potential game-changers in the industry.
“We need more advanced AI agents on the market,“ Ma said. “We need AI agents that can make complex decisions and interact with a bunch of different platforms. Right now, the market is mostly filled with basic and intermediate agents.“
— Cointelegraph (@Cointelegraph) March 19, 2025
The real use cases
Ma admitted that investors often see AI as a buzzword, especially when founders fail to clearly explain why AI is essential to their projects. Saying, “When you see a lot of similar pitches about AI without a clear use case, it feels like it’s just branding,“ she also emphasized AI agents can show specific use cases.
“A lot of AI agents can be a great way to monetize UGC,“ Zhang noted. “A lot of these projects are really community-driven. For example, frameworks like Griffin AI or OpenAI Swarm make workflows more efficient and interact with people in a meaningful way.“
“DeFi and AI make a strong pair,“ Bayan added. “And it’s not just about executing trades — it’s about using AI to monitor positions, run snipers and act when you’re away from your screen.“
Ma mentioned a couple of DeFAI projects that DWF Labs is working on. “HeyAnon combines conversational AI with real-time data aggregation. It helps users manage DeFi operations such as bridging, swapping, staking and borrowing, and analyze trends by pulling insights from platforms such as Twitter, Telegram, Discord and GitHub.”
“Another project we’ve been watching closely is AI16Z, which is redefining traditional fund management models. Their AI agent is like a virtual hedge fund manager — it doesn’t just follow preset rules but analyzes market sentiment, onchain data and trending conditions to make decisions,“ she added.
As speakers noted, there’s also growing attention to how AI agents work together, particularly through agentic workflows and coordination layers. These setups determine whether agents act in sequence or in parallel and how they share data and memory to achieve results. “It’s fascinating to think about agent orchestration,“ Zhang said. “In the future, people believe that more than 90% of a company’s functions could be run by autonomous agents — administration, business development, marketing, accounting — all automated.“
Coordination layers that allow agents to work together effectively will become a key area of interest, according to Zhang: “We need frameworks where multiple agents can organize tasks together to produce a meaningful result.“
She cited the example of Nethermind, an L2 run entirely by autonomous agents: “Each agent registers itself onchain, and transactions are governed by consensus among the agents. Through the Nethermind launchpad, developers can customize agent-run chains for specific use cases, which opens up a lot of possibilities for building fully autonomous systems tailored to different sectors.“
In terms of institutional adoption, Bayan cited regulatory uncertainty and deep-rooted legacy systems as key barriers. He suggested a hybrid approach, where institutions continue to rely on traditional systems while integrating blockchain in certain areas, as a more predictable compromise. “They don’t have to go all in, they can start by letting certain arms experiment with decentralized technologies and build from there,“ he added.
He then pointed to CARV as an example that larger institutions could adopt: “They use blockchain data for benefits and credentials, but rely on offchain machine learning for computation, which is a great bridging model.“
The AI agent future
Not every AI agent project needs to launch with a token right away, according to Zhang. “Some agents are better off being tested in the market first. Getting client feedback and proving the use case should come before designing the tokenomics. A token should improve the ecosystem and governance once the business model is established,“ she stressed.
“I’m not saying AI agents are all about memecoins, but 99% of the time, people treat them that way because they’re launched through community sales,“ Ma continued. “From a venture perspective, we want to see more AI projects built for long-term value with solid products, real revenue and recurring cash flow. We need products with staying power — not just something that peaks at TGE and then disappears. Right now, very few projects are actually achieving sustainable demand,“ she pointed out.
Bayan echoed this sentiment, stressing the importance of ease of use for both end-users and developers: “Users shouldn’t even realize they’re using blockchain or Web3. The next breakthrough moment will be when large Web2 companies start using blockchain-based compute. It needs to feel effortless.“
Zhang pointed to a trend of using AI agents — or bots — to simplify workflows and integrate tightly with platforms that users already use, such as social media: “Functions such as social betting or health information can be easily provided by an AI-powered bot within Twitter or other social media platforms.
I think very soon we’ll see very powerful products supported by AI-driven bots and streamlined AI agents,“ she concluded.
Disclaimer. Cointelegraph does not endorse any content or product on this page. While we aim at providing you with all important information that we could obtain in this sponsored article, readers should do their own research before taking any actions related to the company and carry full responsibility for their decisions, nor can this article be considered as investment advice.