With participation from over 50 projects, how was the quality of Cookie's DeFAI Hackathon?

Original Title: After going through 50+ projects in the Cookie DeFAI Hackathon, here's what I learned
Original Author: Defi0xJeff, Head of Steak Studio
Original Translation: Ashley, BlockBeats

Editor's Note: The author has summarized the insights and reflections after reviewing over 50 Cookie DeFAI Hackathon projects, pointing out the potential and market gap of the DeFAI Agent, especially the rise of Vertical Agents and the shortcomings in researching Agents. The author believes that Cookie, as a data platform, is driving the development of innovative projects and suggests that teams should focus on new use cases rather than replicating existing features. DeFAI is poised to become a significant vertical market in the crypto space and may compete with Web2 Agents in the future.

The following is the original content (reorganized for readability):

After reviewing over 50 Cookie DeFAI Hackathon projects, here are my takeaways (more like feedback / my current view of the Agent market / how projects stand out).

Current State: DeFAI = An Abstraction Layer for Many Developers

Many teams have added an NLP interface to their product front as an afterthought (perhaps because they think DeFAI is equivalent to @HeyAnonai, @griffaindotcom, @orbitcryptoai, @askthehive_ai). In most cases, this is not appropriate—especially when you can only do simple things like using the Cookie API to find the top 5 AI Agent token influencers, identify trending top coins, etc. It's just a mini feature that many top-level abstraction layers already have.

I believe that directly viewing these analyses through the Cookie dashboard would be better than adding a new interface—the functionality is not comprehensive enough.

DeFAI ≠ Abstraction Layer

Instead of replicating existing features, teams should focus on leveraging the Cookie API to unlock new possibilities—driving entirely new use cases and verticals rather than drawing inspiration from existing domains.

Emergence of Vertical Agents

I was amazed by many interesting ideas that emerged in this hackathon—several projects had unique concepts. While many projects are still in early demo/concept stages, they painted an exciting picture of future use cases.

• An Agent to help preserve your legacy—checks on your well-being and takes action to fulfill your wishes if you pass away.

• Utilizing Cookie analysis for investment decision-making and comprehensive research reporting for ETF/index funds. • Agent Security Analysis with Agent Security Scoring.

• A ChatGPT-like product/developer learning hub to help developers understand all things Solana.

• DYOR Layer, tracking analyst/KOL calls, DYOR, and copy trading.

• Framework to enable Agents to sign contracts, enabling complex Agent-to-Agent or Agent-to-Human interactions (unsecured loans, hiring agreements, alliances/coordination).

• Personalized DeFAI Agent—AI companion that adjusts its behavior/visuals based on market dynamics.

More teams are launching niche-specific Agents rather than just a "trading Agent" or AI-driven dashboard/research Agent. Introducing vertical Agents makes them easier to distinguish from general Agents.

Trading Agents already have key players. Although this field is still in its infancy, standing out remains challenging, especially in the early stages. Focusing on vertical Agents would be better.

Many may think @HeyTracyAI is the flagship Agent on Solana by @virtuals_io, not useful, unable to make you money. In reality, an Agent built like a real business—solving real issues—will perform better in the long run. The sports market is a huge TAM. Look beyond Web3 fields. (Not promoting Tracy, just putting forth a point on vertical Agents.)

Conclusion: Niche-specific vertical Agents address real problems, creating unique use cases, while general Agents struggle to stand out.

Lack of Suitable Research Agents

While vertical Agents are carving out unique niche markets, another significant gap in this field is a suitable Research Agent.

The key word here is "suitable." Currently, there is no research on whether an Agent can replace human information synthesis and reasoning. This applies not only to the Cookie DeFAI Hackathon project but also to the general case of a Web3 AI Agent.

Most AI Agents today simply aggregate data without synthesizing insights like humans. Analyzing data through traditional dashboards such as @cookiedotfun, @GoatIndexAI, @Decentralisedco, and using Grok is still better than having an AI Agent "feed" insights to a Web3 AI Agent.

Despite many abstraction layers and teams focusing on enhancing research capabilities, there is still a noticeable gap here. Whoever can first break through this will have a significant advantage.

Cookie DeFAI Hackathon Project

Most hackathon projects are still in the early stages of development, and many projects have not yet been deployed. As this is a pure DeFAI hackathon (as you can see, DeFAI is the best-performing category in AI Agent), many high-quality projects and tokens will emerge from this event.

As mentioned in the second part, many projects will provide new use cases beyond what we currently understand as DeFAI applications.

As the AI Agent field continues to evolve, Agents can fill more gaps—such as B2A (Business to Agent) surpassing B2B and B2C.

The next wave of DeFAI projects will not only enhance existing use cases—they will create entirely new ones.

Cookie as a Data Support and Distribution Channel for Agents

Unlike relying on a launchpad to highlight a unique Agent token, Cookie empowers Agents and teams by providing on-chain and off-chain AI Agent data support, enabling new and exciting use cases.

At the same time, Cookie's dashboard has been used by over 240,000 MAU, these users are deeply rooted in the field. The gems found on the Cookie dashboard and in the Cookie hackathons are like discovering a new gem on Virtuals.

Cookie has proven to be a powerful Agent distribution channel. The more Agents leverage this, the faster the ecosystem matures.

Conclusion

This hackathon felt similar to the Solana AI hackathons, but arguably better—because it was a pure DeFAI play.

DeFAI is not just another AI trend—it has the potential to become the most promising Agent vertical in the crypto space. This hackathon proved that.

I lean towards DeFAI, believing it to be a native Agent use case in crypto, able to develop as a standalone vertical and compete with Web2 Agents.

「Original Article Link」