Written by: Portal Lab

"Have you raised lobsters?" Recently, Web3ers greet each other, and it's likely that eight or nine out of ten will say this.

At the beginning of 2026, after a robotic performance at the Chinese New Year Gala shocked everyone, a new generation of AI Agents represented by OpenClaw became the new toy among tech enthusiasts. Some use AI for customer service, some use AI to write code, and some have even started attempting to simulate an entire set of "digital employees" using Agents. A concept that has been frequently mentioned on various internet platforms lately, "one-person company," refers to an individual using an AI workflow to accomplish what used to require a small team.

Web3 hasn't been idle either. Recently, if you take a closer look at industry media, you'll find that many projects have started to focus on AI Agents. Some are researching how Agents can directly invoke on-chain assets or contracts, some are working on payment, identity, or financial infrastructure for Agents, and others are discussing the "Agent economic system," allowing AI to participate in the network like users, with some even beginning to shout the new slogan "Web4.0."

Seeing this, it will actually evoke a sense of familiarity.

It is said that the fashion industry is cyclical, and it seems the tech industry (or crypto circle) is no different. Remember the bear market that started in 2022 when ChatGPT exploded in popularity overnight? AI suddenly became the talk of the town for everyone. Certainly, the Web3 circle was also busy, quickly producing a slew of new concepts like AI Agents, AI traders, and automated strategies, as if any mention of AI could lead to a new story. But this excitement did not last long. Once the crypto market began to rise again, everyone's attention quickly returned to Crypto itself.

And now, in the second half of 2025, the crypto market shows signs of a bear trend again, prompting Web3 to look for new concepts to latch onto.

However, in the view of Portal Labs, the problem lies precisely here. When a narrative starts to become popular, many Web3 startup teams are not actually making technical and business judgments but are instead making narrative judgments: whatever concept is hot, they pursue that, which leads to their downfall.

Many teams realize when genuinely advancing projects that concepts can be quickly established, but products are difficult to implement. Where are the users? What are the specific scenarios? What will sustain ongoing billing? Can investment be secured? These issues often emerge gradually only after the project has been underway for some time.

By the time interest wanes, what remains in the market is often a pile of projects that have not been validated. Some products stall at the demo stage, some barely launch but fail to attract users, and some simply disappear along with the narrative. In the short term, it may seem like a new track has been opened, but looking back after some time, there is not much left that is substantive.

Thus, the dilemma arises: whether to continue to delve into Crypto or pivot to AI. Choosing the former means the market is poor and investment may not yield returns; choosing the latter offers no guarantees. The technical thresholds, talent structures, and competitive environments of AI differ from those of Web3. The tech stacks, product experiences, and community resources many teams have accumulated over the past few years are essentially rooted in the Crypto framework. Once they completely shift to AI, it is akin to entering a completely unfamiliar domain. From model capabilities to data resources to engineering teams, nearly everything would need to be rebuilt.

More realistically, the AI track is already very crowded. Whether it’s large model companies, traditional internet firms, or numerous startup teams, immense resources have been invested in this field. For a startup team originally focused on Web3, if they enter this market merely due to narrative shifts, it is easy to find themselves without technological advantages or industry resources.

In fact, for many Web3 startup teams, there is another practical path. They don’t necessarily have to transform into AI; instead, they can continue to pursue their Web3 path while thinking about what capabilities Crypto can contribute to the AI system.

If you take a closer look at the current wave of AI development, you'll discover that many key aspects are still not fully resolved.

The most typical example is data. Models are becoming increasingly powerful, but where does the training data come from, is the data trustworthy and compliant, particularly how can AI Agents achieve 1v1 customization? These questions have yet to be addressed by a solid mechanism. For AI dependent on large-scale data training, this poses a long-standing foundational issue.

Another example is identity and cooperation. When AI Agents begin participating in task execution, automated trades, or even operational decisions, they themselves also need identity, permissions, and collaboration rules. Who can invoke a specific Agent? How do Agents divide labor? How is settlement done after executing tasks? These questions essentially concern identity and value distribution in open networks.

Then there is the payment issue. Once AI Agents autonomously invoke services, obtain data, or execute tasks within the network, it means they require a small payment system capable of automatic settlement. However, achieving such a payment structure is quite challenging within traditional internet systems.

These seem to be issues for AI, but many solutions actually exist within the technical framework of Crypto. Whether it’s data incentive networks, on-chain identity frameworks, or open payment networks, these have been directions that Web3 has been exploring over the past few years.

If Web3 startup teams genuinely intend to experiment in these directions, there are several things they must consider first.

The first thing to look at is the technical capabilities of the team itself. There is a significant variance in technical accumulations among different Web3 projects. Some teams excel in developing on-chain protocols, some have long been engaged in data networks, and others lean more toward application-layer products. If the team has spent the past few years working on data-related infrastructures, such as data collection, data extraction, or data markets, then extending into the data layer surrounding AI will be relatively natural, such as creating data contribution networks, verifiable data sources, or providing incentivized data markets for models. If the team originally focused on on-chain protocols or infrastructure, they might consider working on the operational environment for AI Agents, such as on-chain identity, permission management, task execution protocols, or providing automatic settlement and payment capabilities for Agents. For those already developing application-layer products, like trading tools, content platforms, community products, or consumption applications, AI is better as a capability layer integrated into the existing product system. For instance, using AI to enhance data analysis capabilities, automate operational processes, or let Agents complete originally manual tasks.

The second thing to consider is whether there are real business scenarios. Many AI projects disappear quickly not due to lack of technology, but simply because there was no clear use scenario from the start. Concepts can sound very appealing, but where are the actual users who need this product, why would they use it, and why would they be willing to pay for it? These questions are often not answered seriously. Some concepts are heavily discussed in the industry, like "AI + Web3," "Agent economic system," and "AI traders," which sound grand, but when pressed further, the stable user base that actually exists is quite small. Conversely, some needs that seem less "sexy," such as data processing, automated operations, information filtering, or task execution, have existed in real business contexts for a long time. For this reason, when judging whether to enter a particular AI direction, rather than looking at whether the concept is popular, it is more useful to examine the scenario itself: is this scenario a long-existing business problem, is there someone already paying for it, and can AI genuinely improve efficiency in this segment? If these conditions are met, then that direction is more likely to transform from narrative to product.

Additionally, it is essential to assess whether the Web3 startup teams have the resources that can genuinely engage with these aspects.

As previously mentioned, directions like data, identity, and payment are not merely technical issues; they are network resource issues.

For instance, regarding data networks, if the team lacks a stable data source or a user base capable of continuously contributing data, then even if the technology is developed, it will be difficult to form an actual network effect. Similarly, if they wish to create an identity system or cooperation network for AI Agents, they need real developers, applications, or Agents to participate; otherwise, the protocol itself will struggle to form an ecosystem. The logic for payment and settlement systems is similar. Once AI Agents start invoking services, obtaining data, or executing tasks on the network, small payments will become very frequent. However, this payment network only holds significance when a large number of Agents and services exist simultaneously; otherwise, it remains merely a technical module.

Thus, for many Web3 teams, the critical evaluation should not be whether "there is technical space in this direction," but rather whether they can become part of this network. Whether the team already has data sources, developer ecosystems, or application scenarios often decides if a project can genuinely enter the foundational layer of AI infrastructure, rather than merely remaining at the conceptual level.