Editor’s note: The author summarized the gains and reflections after reviewing more than 50 Cookie DeFAI hackathon projects, pointing out the potential and market gaps of DeFAI Agent, especially the rise of vertical Agents and the insufficient research on Agents. The author believes that as a data platform, Cookie is driving the development of innovative projects and suggests that teams should focus on new use cases rather than replicating existing functions. DeFAI is expected to become an important vertical market in the encryption field, capable of competing with Web2 Agents in the future.

The following is the original content (the original content has been edited for ease of reading and comprehension):

After browsing through more than 50 Cookie DeFAI hackathon projects, my gains (this is more like feedback / my current view on the Agent market / how the project stands out).

Current situation: DeFAI = the abstraction layer for many developers

Many teams add NLP interfaces in front of their products (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, such as using the Cookie API to find the influence of the top 5 AI Agent tokens, and looking for promising top coins. This is just a mini feature that many top abstraction layers already have.

I think it would be better to directly use the Cookie dashboard to view these analytics, rather than adding a new interface - its functionality is not comprehensive enough.

DeFAI ≠ Abstraction Layer

Rather than duplicating existing functionality, teams should focus on leveraging the Cookie API to unlock new possibilities – driving entirely new use cases and verticals, rather than taking inspiration from existing ones.

The Birth of Vertical Agent

I was amazed at the many interesting ideas that came out of this hackathon – several projects had unique concepts. While many projects are still in the early demo/ideation stages, they paint an exciting picture of future use cases.

• An Agent to help preserve your legacy - check if you are safe, and if you pass away, it will take action to fulfill your wishes.

• ETFs/index funds that utilize cookie analysis for investment decisions and comprehensive research reports. • Agent security analysis and agent security score.

• Similar to the product/developer learning center of ChatGPT, it helps developers understand all the information about Solana.

• DYOR layer, track analyst/KOL calls, DYOR and copy trading.

• Enable the framework for Agents to sign contracts, enabling complex interactions between Agents or between Agents and humans (unsecured loans, employment agreements, alliances/coordinations).

• Personalized+DeFAI Agent-AI companion, will adjust its behavior/visual effects according to market dynamics.

More and more teams are launching niche Agents, not just ‘Trading Agents’ or AI-driven dashboards/research Agents. Introducing vertical Agents makes it easier to distinguish them from general Agents.

The trading agent already has a top player. Even though the field is still in its infancy, it’s still hard to stand out – especially in the early stages. It would be better to focus on vertical agents.

Many people may think that @HeyTracyAI is @virtuals_io’s flagship agent on Solana, which is useless and can’t help you make money. In fact, an agent that is built like a real business – solving real problems – will perform better in the long run. The sports market is a huge Total Address Market (TAM). Look beyond Web3. (Not pushing Tracy, just making a point about the vertical agent.) ）

Conclusion: Vertical Agents in segmented fields solve practical problems, create unique use cases, while General Agents struggle to stand out.

Lack of suitable research agents

Although the vertical Agent is opening up unique niche markets, another major gap in this field is suitable research Agent.

The key here is ‘suitable.’ Currently, there is no research on whether Agents can replace human information synthesis and reasoning. This applies not only to projects like the Cookie DeFAI hackathon, but also to the general case of Web3 AI Agents.

Most AI Agents now just aggregate data, but do not have the comprehensive insight like humans. Analyzing data through traditional dashboards, such as @cookiedotfun, @GoatIndexAI, @Decentralisedco, and using Grok, is still better than letting AI Agents ‘feed’ Web3 AI Agents ‘insight’.

Despite the presence of many abstract layers and teams focusing on enhancing research capabilities, there are still significant gaps here. Whoever can first break through this point will have a crucial advantage.

Cookie DeFAI Hackathon Project

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

As mentioned in the second part, many projects will provide new use cases beyond our current understanding of DeFAI applications.

With the continuous development of AI Agent as a field, the Agent can fill more gaps - for example, B2A (Business to Agent) surpasses B2B and B2C.

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

Cookies serve as an agent’s data support and distribution channel

Unlike relying on launchpads to highlight unique Agent tokens, cookies empower Agents and teams by providing a support that tracks on-chain and off-chain AI Agent data – which enables new and interesting use cases.

At the same time, Cookie’s dashboard has been used by over 240,000 MAU, and these users are deeply cultivating this field. The gems discovered on the Cookie dashboard and Cookie hackathon are like finding a new gem on Virtuals.

Cookies have proven to be a powerful Agent distribution channel. The more Agents take advantage of this, the faster the ecosystem will mature.

Conclusion

This hackathon feels similar to the Solana AI hackathon, but you could say it’s even better - because it’s pure DeFAI.

DeFAI isn’t just another AI trend – it has the potential to become the most promising agent vertical in the encryption space. This hackathon proves it.

I tend to favor DeFAI, considering it as a use case native to encryption, capable of developing as an independent vertical domain and competing with Web2 Agent.

“Original Text Link”