Original author: Defi0xJeff, head of steak studio
Compiled by: zhouzhou, BlockBeats

Editor’s note:This article discusses the correct way to build cryptographic AI agents. First of all, it emphasizes that agents should be supplements to core products and enhance product value by changing user paths, rather than simply relying on tokens. Entrepreneurs should focus on practical problem solving, create sustainable core products, and then promote and make profits through agents and tokens.

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

The market has experienced retractions again and again, and liquidity has gradually become thinner and thinner. The maximum market value of the new agent successfully launched recently is approximately US$10 million.

By “success” I mean that the product has product-to-market fit (PMF), provides value to actual users, and has started (or is about to start) generating revenue. This is completely different from what was happening three or four months ago, when agents with PMF could have a maximum market value of more than $100 million, especially if they were positioned as agent + framework/launch platform tokens. For example, AVA, as a 3D agent, not only as the agent itself, but also obtains value from launching the platform and supporting projects through its audio-visual layer.

Old Manual: Agents as Framework

The approach was to launch an agent to demonstrate its capabilities, attract developers to want to build their own agent, and require those developers to hold/destroy/pay agent tokens to access the framework. The problem is that the crypto community places exorbitant premiums on framework tokens, and these “framework agents” often lack differentiation. In many cases, they don’t even have a product-they’re just babbling on Twitter, hoping that token prices will rise.

The first version of proxy included Conversation Proxy itself as a product, which is unique in the crypto space because we put more emphasis on community building-similar to founder led marketing (founders use chatter to gain attention). Getting agents to chatter about your project seems like a good idea to add attention to your project-it worked well when first launched in November 2024 and lasted for a month. But now, with 420,690 agents babbling non-stop, most agents seem immature, repetitive and, frankly, annoying.

New Handbook: Agents as Business

Here’s how you should think about launching an agent-launching an agent means you’ll be running a startup and managing up to three products simultaneously:

1. Core products (actual business)

Your core product should solve practical problems. It should not be just a conversation agent, but a real product.

Example:

·Improve prediction models for sports betting odds to help users win more in sports betting (e.g., crypto community AskBillyBets).

·Crypto asset prediction models that better perform transactions, reduce temporary losses, and maximize returns to liquidity providers (e.g., crypto community Cod3xOrg, crypto community gizatechxyz, crypto community Almanak).

·Aggregating insights from top alpha sources such as Cookie, Kaito, Nansen, Messari, Aixbt, CG, Dexscreener, and Bubblemaps to help AI agents in investment decisions research search engines (there is no team yet to solve this problem-we need an AI agent similar to Perplexity).

Core products should be the top priority for each team before launching tokens. You need to make sure there is real demand in the market and users are willing to pay for it. Otherwise, you will fall into a “death valley” in the crypto world, with consequences that may be more serious than traditional startups:

·High operating costs.

·Use tokens to pay customer acquisition costs (CAC).

If your tokens plummet, it will become a curse. Most people won’t care about your project, no matter how strong your core product is or how much progress you make.

Instead of relying on token incentives, focus on attracting customers through products. Find a profit model that balances growth with revenue generation.

The manual from the crypto community KaitoAI is a good case study:

·They built an enterprise product-a cryptographic search engine that focuses on social/emotion/narrative and charged users, projects and ecosystems to provide real value.

·They launched the Mindshare Dashboard as a standard for tracking narratives and trends.

·They stepped up their efforts to launch the Yapper rankings and let KOL spontaneously share it as a status symbol.

·They further launched NFT WL and KAITO airdrops to stimulate Twitter interaction through practical rewards.

Likewise, communication is very important. Many projects have strong products but poor communication skills. If no one knows what you’re doing, no one will care.

2. Tokens (alignment tool)

We have shifted from “venture investment coins” to “fair issuance”, celebrating high-liquidity, low-FDV tokens. But fair issuance is not entirely fair-every token strategy has its own trade-off.

If you launch proxy tokens with high liquidity and low FDV structures, you will not be able to raise funds from venture capitalists and angel investors (because the valuation is too low). However, you can use tokens as a marketing tool to initiate thought sharing.

Many teams launch two types of tokens:

·Proxy tokens → Start thinking sharing.

·Ecological Tokens → Raise funds from venture capitalists and angel investors at higher valuations.

But this creates a mismatch of expectations-the community expects airdrops, and when the eco-token is launched, capital is transferred from the proxy token to the eco-token, causing the proxy token price to collapse. Managing core products + proxy tokens + ecological tokens while ensuring the value accumulation of each token is very complex.

In an ideal world, there would be a token to accumulate all the value from the core product. Historically, projects that generated revenue and returned it to tokens (through repurchase or revenue distribution) have survived in the long term.

Tokens should be supplements to core products, not necessary conditions. For an in-depth understanding of proxy token strategies, please check out the crypto community VaderResearch’s analysis of the proxy token manual of the crypto community virtualsio.

3. Agent (Supplementary Products)

“Proxy” refers to a dialogue proxy built using frameworks like ElizaOS, G.A.M. E, ARC, Pippin, etc. Although these agents integrate on-chain/off-chain capabilities, they should be complementary to the core product.

Agents should enhance the value of core products by changing user paths:

Instead of letting users proactively find and use your product, let an agent bring it to them.

This can mean:

·Display products directly on Twitter via text/video.

·Use agents as AI companions to change the way users interact (similar to ChatGPT abstraction).

·The agent acts as the interface itself and performs tasks behind it.

Of course, there are exceptions. Aixbt agent is an example-it provides real-time social and emotional analysis from Twitter, allowing users to get real-time Alpha signals before others do. Aixbt has become the NO.1 KOL on CT by continuing to provide Alpha and demonstrating the capabilities of the terminal. In this case, the agent itself is the product.

However, this model is very difficult to replicate. Most projects should focus first on strengthening core products.

A product-first success story is cookiedotfun:

·Start with a free AI agent dashboard to attract users.

·Move to a paid value-added model and unlock advanced insights by locking in cookies.

·Monetization by providing APIs for projects and agents.

·Launch agentcookiefun to bring insights directly to Twitter.

summary

In 2020-21, launching tokens requires mastery of Solidity. But now platforms like pumpdotfun make it easy to tokenize anything.

This changes the way of thinking-people are no longer focused on building real products and are launching tokens directly. This method is “garbage comes in, garbage comes out”, and capital will be quickly transferred to the next “garbage”.

We need to change that.

In order to build sustainable projects, agency projects should be treated like startups. Instead of looking for funding among CT, VC and angel investors, build projects with long-term value-not for the next six months, but for the next six years.

Innovate, solve practical problems, and create real businesses-rather than just building the next speculative token farm.

The future of cryptographic AI agents depends on this.

“Original link”