Article reprint source: AIcore

Original source: What AI invests in

Image source: Generated by Unbounded AI‌

Overall, I don’t think there will be disruptive ecological changes or so-called “killer” applications known to the public in China in 2024. Some self-media like to compare it to the “XX moment” of mobile Internet, but the time scale in the AI ​​era will be different. It may take three to five years or even longer to compare it to a previous release.

Nevertheless, the real user needs that AI can meet in many areas will become clear. We will not know who will ultimately win in 2024, but we can be sure that certain segments will be able to achieve sufficient cash flow.

The fragility of OpenAI's management structure has allowed other large-model companies to see the opportunity to defeat it from a commercial rather than a technical perspective. Some companies are willing to adopt more radical innovations in their business models, thereby providing differentiated options for application products.

The suitable entrepreneurial fields for direct 2C will still be scenarios with a high tolerance rate, such as entertainment, office, social networking, etc. In the fields with a low tolerance rate, such as education, medical care and finance, it is a better choice to adopt the B2B2C approach and give the responsibility of judgment to professionals rather than users.

The following brief guesses include AI companionship and virtual social interaction, AI+games, AI+content creation (non-games), AI+education, AI+personal productivity tools, and AI+personal life assistant. Due to space limitations, each is only a few sentences long. I will try to write an article for each direction before I am slapped in the face.

01 AI companionship and virtual social interaction

In the new year, it will still be one of the most popular directions for 2C applications. Products with atypical AI vertical user groups as the differentiation focus will increase and grow rapidly. General virtual character platforms and products with high-frequency AI users will have a hard time maintaining traffic. The means will include richer multimodal interactions, high integration with hardware, and attracting high-quality PGC partners with their own IPs to settle in, etc., in order to raise the financial threshold for competition in this field. However, the overall domestic competition situation will be one or two steps slower than overseas.

Platform products will begin to earn revenue from the consumer side. The most likely model to achieve phased results in China is to refer to the online literature model, where users subscribe to and reward UGC, and the platform and the author share the profits. Non-platform products will be closer to the business model of games, but they will not be able to gain an advantage in the competition with ordinary game substitutes.

As a relatively new demand brought about by new technologies, the focus of companion products is on how to find the right people and stimulate specific needs and payment points. The team's sensitivity and operational capabilities in this regard are more worthy of attention. At least in 2024, product design details are not the decisive factor.

02 AI+Games

In a sense, games are also a type of companionship product, and virtual companionship is also a type of AI game. In 2024, the boundaries between the two will become increasingly blurred.

Large companies will actively use AI as a production tool, but integrating AI into gameplay will not become a common phenomenon. After all, changes to the planning of ongoing projects must be made with caution, and the establishment and development of new trial games takes time. Any risk that may destroy the experience of existing players is not worth taking.

New startup teams have no baggage, but will focus on light games. It is easy to make money but difficult to make money, not to mention the hurdle of obtaining a game license. Although going overseas is good, there will be no AI game products that are truly worth going overseas in 2024.

AI+games are first and foremost games. Whether starting a business or investing, it is best to ask yourself: "Does anyone on our team really understand game planning? Or are we just using AI to find a way out and making choices by elimination?" "Our organization has not invested in the game industry for ten years. Is it really enough to judge whether a game team can succeed based on a total of thirty years of player experience?"

03 AI+content creation (non-game)

The transition from toys to productivity has already occurred rapidly in 2023 and will become a common understanding in 2024, regardless of whether users actively pay attention to AI and whether existing practitioners are willing to accept it.

The cross-use of AI content creation will be greatly developed in 2024, and independent plot content creators will begin to emerge in large numbers. The next big company will begin to be born in the emerging content publishing platform that is closely integrated with the tools, and will first eat up the market share of small online literature companies and comics companies that stick to the old ways.

The most anticipated AI drama videos are not yet realized, but various video formats such as ghost animals, live shows, splicing, face-changing, etc. are opportunities. If the existing UGC platform manufacturers cannot or do not want to grasp it, there will always be people who will act as traffic porters.

04 AI+Education

A large number of AI education products for students will be launched, trying to prove that "personalized learning" or "improving learning motivation" can be achieved through AI technology, but at least in 2024, the effect will not be much better than the existing education apps, and may even lead to doubts about cutting parents' leeks. Existing education hardware manufacturers will gain an advantage because parents do not need to pay for AI products separately.

Another group of products will avoid the primary and secondary school stage, which has the lowest tolerance rate, and look for opportunities in various groups such as preschool, higher education, vocational education, and quality education. In general, these markets are much more tolerant of AI gimmicks, and how to use AI to increase the fun under existing technical conditions has a lot of room for product managers to play.

For applications with teachers as their user base, B2B2C that directly pushes them to teachers and that uses schools as public relations targets will form completely different product forms. The former is more suitable to be classified as a personal productivity tool, while the latter can be integrated into the education system to become a true AI education product and show certain competitiveness in 2024 - but it is very likely that the main battlefield will be overseas.

05 AI+Personal Productivity Tools

Startup teams that develop various tools to improve personal productivity, such as office, compilation, and editing, will face strong competitive pressure from large companies and large models. A large number of cannon fodder will appear in 2024, even for some products that have already accumulated a wave of users in advance.

However, this does not prevent more early-stage teams from entering these fields, because many people are confident that they have a more accurate grasp of "easy to use" than others, and can successfully transform into products with higher ceilings after gathering enough users.

As the number of companies accumulates, a very small number of companies will have the probability of escaping the blockade of large companies, being incorporated and completing entrepreneurial exit, but there will not be mergers and acquisitions of a sufficient scale in 2024.

06 AI+Personal Life Assistant

The conceptual description, abstract definition and endless imagination of Copilot or personal Agent will no longer be the focus of the hustle and bustle. Everyone will remember that applications are made for application. It is whatever it is used to achieve. It doesn't matter if it is called a kitten or a lamb. . In 2024, AI will realize some practical functions in personal life. Rather than forming a viable product independently, it is more likely to be hidden in some existing websites and apps, allowing some users to try to get suggestions, reduce manual operations, etc. Get results with just one click.

Although many Internet giants like to say that "all applications can be reconstructed by AI", there will inevitably be differences in the speed of implementation. The right direction in the long run may be the wrong entrepreneurial choice in the short term. The starting point of a personal AI assistant is to implement high-frequency and rigid but error-prone functions first, or to implement low-frequency and non-rigid but forgivable functions first. There will be different views under different operating strategies, and this is just one example of a large number of selection dimensions.

Throughout 2024, practitioners may be looking for answers to the question "What kind of Agent should and can be?" However, instead of staying on PPTs and official accounts, they will improve and correct their cognition in the cruel life-and-death test of market practice.

07 Conclusion

This is a conservative forecast that lacks imagination and represents our minimum expectation for the development of AI applications on the C-end next year. Although we have always said that trial and error are inevitable in the development process, in the current environment of tight funding, it is important to measure one's own capabilities as much as possible and reduce the consumption caused by trial and error.

It is better to be too cautious and have to catch up than to be too optimistic and end up reflecting and complaining.