Tokens are the best tool to solve the "cold start" problem, which is more responsible for the failure of creative projects than any other factor.
Author: Grant blocmates, Founder of blocmates
Translation by: Deep Tide TechFlow
Thanks to @brextonpham, @devlordone, and @yashhsm for the discussions over the past few weeks that helped clarify my thoughts on this topic.
Recently, a group of entrepreneurs from the Y Combinator cohort has frequently been asked, "What percentage of your codebase is generated by AI?" The results showed that one in four (25%) teams reported that 95% of their code was generated by AI.
Andrej Karpathy, former AI director at Tesla and founding member of OpenAI, coined a controversial yet fitting term—"Vibe Coding."
In simple terms, this means that ordinary people like you and me can generate code directly through text prompts. This trend has inspired countless "Ideas Guys," who have postponed their job search plans to try to create the next Facebook.
Independent entrepreneur Pieter Levels recently showcased the potential of AI in application development. He developed a flight simulation game that is almost entirely AI-generated, which has already reached a recurring revenue of $85,000 per month solely from built-in advertisements.
However, this approach still has many issues, such as some bugs requiring developers to have basic technical knowledge to understand what went wrong after "madly typing on the keyboard."
As a result, some real developers have started offering consulting and debugging services to help the next generation of AI-dependent programmers solve problems (yes, this front-end code is a product of "Vibe Coding").
As these cutting-edge models continue to improve, their ability to generate code from natural language (text or voice) will become increasingly efficient.
Dario Amodei, CEO of Anthropic, recently stated that within "three to six months," AI will be able to complete 90% of coding tasks. And within twelve months, AI will "essentially complete all coding".
What would the world look like if AI programming became the norm?
Who will be the losers?
The cost of application development will approach zero, and those SaaS services you don't want to subscribe to can be replicated at a very low cost.
If a SaaS company has not established a strong moat through network effects, hard-to-replicate products, or strict intellectual property protection, its fate is predictable. Companies like Docusign and Typeform are likely to be eliminated.
Moreover, businesses can develop native products more efficiently through MCP (Model Context Protocol), which can seamlessly connect the team's workflows, knowledge bases, and existing databases.
I believe that "efficiency employees" who know how to leverage AI for development may currently be one of the most sought-after roles in small and medium-sized enterprises (SMEs).
With the launch of some powerful small models, businesses can even run all systems on local machines or models, which is good news for privacy and operational security (OpSec).
Teams that still believe "as long as the product is good, users will come" will also be eliminated, as smarter teams will clone their products and market them with better marketing strategies.
Who will be the winners?
As code generation models evolve, the technical barrier will gradually lower, making the competitive environment fairer. Teams that can dominate user minds and attract attention will perform exceptionally well.
Excellent marketing comes from a deep understanding of the core audience, which often stems from the team itself being part of the target user group. If the basic functionalities of all products are similar, applications will resemble fashion items, becoming objects of people's bragging.
For example: "I use Claude because I know ChatGPT is for those ordinary people who don't understand the field."
To make an application something people are willing to recommend, it must be "cool" enough. Brand building, positioning, and cultural sensitivity must be in place. This is precisely the area where large language models (LLMs) will struggle to reach in the short term. The reason is that human complexity far exceeds AI's imagination.
Sub-cultures are particularly difficult to understand from the outside. What impresses one person may hold no appeal for another.
In the future, we may see a world where every user can enjoy a personalized user journey, interface design, and even brand experience based on their preferences, all accomplished without the user being aware of it.
Roles with exceptional aesthetics and keen insights into design trends—such as graphic designers, creative directors, marketers, and UI/UX experts—will continue to play important roles in organizations.
These positions require "taste" and the ability to anticipate trends, which is always a capability that models cannot replace. After all, at its core, a model is just a "text prediction machine."
App-Coin Theory
I want to clarify that I am very aware that the term "App-Coin" sounds a bit silly, but the ironic effect in written expression is often not obvious enough.
Unless you are already a shareholder in companies like Anthropic, Cursor, OpenAI, Perplexity, or have participated in application-level investments like Lovable, Replit, LindyAI, you may only be able to "be happily poor."
If you are a super gambler and still believe that cryptocurrency is the future, then some teams I have recently come across are building around this future.
Their idea is "App-Coins." These tokens could symbolize founders or small development teams, even carrying the hopes of future application development like "Meme coins."
Tokens are the best tool to solve the "cold start" problem, which is more responsible for the failure of creative projects than any other factor.
For founders and small teams that cannot attract the initial 10-100 users, leveraging tokens and speculation can effectively address this issue.
It is undeniable that humans love speculation. They also want the process of speculation to be fair. The idea of finding the next genius developer and betting on their latest idea makes sense.
Did you see this? Cal AI can estimate calorie content with a margin of error of plus or minus 10% by taking photos of food. Sounds like the first business idea everyone would think of, right?
But behind it is a group of four teenagers who started with a feature from MyFitnessPal and built a business with an annual revenue of $20 million. How did they do it? By deeply understanding brand building, positioning, and accurately targeting marketing audiences.
They heavily reinvested their monthly income into influencer marketing (which is indeed an effective way to discover potential users in the fitness industry) and positioned the product as the choice for ordinary users: simple, refreshing, and fun.
At the same time, they made it clear that if you are preparing for the "Mr. Olympia" competition, this product may not be suitable for you.
How many times has this product failed? If I had to guess, probably hundreds of times.
What we will see next is that applications will be launched alongside tokens. This approach, for better or worse, at least allows them to gain some attention in the early stages.
Will it be abused? Undoubtedly.
Will there be projects with no real content? Most likely, yes.
Will it still happen? Bet on it, it definitely will.
What excites me is that the emergence of code generation models and "Vibe Coding" allows anyone, anywhere, at any time, to bring an idea to market. (Although to be honest, the name "Vibe Coding" is already making me a bit uncomfortable.)
Those introverts who don't like to let carrots touch peas can also bring products to market without making eye contact or talking to anyone. This is good news.
Teams I am watching:
@KaitoAI: In an era where anyone can easily launch a product, mindshare and attention are the top priorities.
@devfunpump: A "Vibe Coding" platform that helps independent founders and small teams go from 0 to 1 on Solana and tokenize their ideas. I love this team and hope they succeed. Special mention to @buidldao_.
@tryoharaaAI: A Base-based project that allows you to go from idea to product with just a few clicks, with outstanding UI/UX. The team is reliable, and Brexton is a genius.
@sendAIfun: Initially starting with smart agents, they are now developing an application environment that anyone can use. Here is their recent interview.
@0xtarobase: Currently quite low-key, but I believe they are incubated by Alliance.
These are some teams I can think of, but I believe that in the future, some talented independent creators will emerge from obscurity and shock us. I even think we will see a billion-dollar project launched by independent founders within the next 12 months.
Summary
AI allows even those who cannot code to push their ideas into production. Tokenization brings speculative opportunities, significantly increasing developers' chances of attracting initial users. An application boom is on the horizon.
If the fantasy I envision comes true, it will be one of the largest wealth creations we have ever seen. Not only will it allow founders to create immense wealth, but it will also provide ordinary users the opportunity to invest in ideas before others.
We have already seen a significant demand for a fair competitive environment that gives retail investors more opportunities. Pumpfun is a response to low liquidity, high FDV (Fully Diluted Valuation) projects, but it is still not the answer.
Investing in "garage band devs" who know how to market their products is a new opportunity that lies ahead for everyone.
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
