# Centum Product Plan FY2 S3 V1

Also see [Centum Product Plan FY2 S2 V1](../FY2%20S2%20(24%2001%2001-24%2004%2001)/Centum%20Product%20Plan%20FY2%20S2%20V1%204ae430878e4a4cee92670b212a920ab6.md). 

# Centum

NOTE: Like any other existing content apps, such as TikTok, Instagram, or YouTube, Centum is easier to understand with a demo rather than with a verbal description. So try it! But here, we take a shot at explaining it verbally. 

Centum is the contents app for the intelligentsia: people who love to learn. 

Centum is deep like Wikipedia and fun like TikTok. 

- Like Wikipedia, Centum is a library of text documents that dive deep into the topics of human interest.
- Like TikTok, Centum presents contents in a way that is easy and fun to consume based on present context and past interactions.

Centum is a great solution to curiosity and boredom, especially for TikTok haters who simply cannot enjoy those short shallow videos. Start the journey by opening a document on a topic that you searched or got recommended, then let your mind wander joyfully by following links in the words of the opened document, in doing so growing your understanding of the world. 

# Product

## Fun like TikTok. Deep like Wikipedia.

For people like me, neither TikTok nor Wikipedia is the perfect solution for fulfilling curiosity or finding fun. 

The cost of automatically understanding and therefore searching for videos is prohibitively expensive, today and in the foreseeable future, so TikTok remains an ineffective solution for anyone who is curious about a specific topic and for anyone who likes to scan fast and skip to the most interesting part. 

Articles on Wikipedia and other such long form text contents on sites like Reddit and small blogs are the current best solution that enables people to dive deep into specific topics and skip to most interesting parts. If text is converted to audio, the person can even keep learning while doing other productive tasks such as running for fitness. 

However, those current best solutions were invented decades ago and are far from what is possible with currently available technology. 

Imagine a Wikipedia made by TikTok. Such solution will be deep like Wikipedia but also fun like TikTok. The content consumer will be able to either search for a specific topic, like on Wikipedia, or get recommended interesting topics based on present context and past interactions such as where they are, what time it is, and whether they skipped the wine document and instead spent a long time reading about coffee. Documents will also be tailored to the person and the context, such as whether the person reached the Stanford document from the admissions rate table in the Harvard document or from the travel destinations section in the California document. The producer of such solution will also understand the importance of making infinite scroll fun and easy such that content consumers can be guided to learn a lot about the world, especially from deep long contents rather than from short shallow contents. 

That is Centum. 

## Wireframe

This is a wireframe that has been made rapidly using Notion in order to show what Centum may look like.

![Centum Notion.png](Centum%20Product%20Plan%20FY2%20S3%20V1/Centum_Notion.png)

Clicking on a blue link will open a document that appears trivially similar to a document on Wikipedia but with sections that are ranked and worded specifically for the consumer.

# Enterprise

- Skipping Repeated Discussions
    
    We do not discuss following topics in this product plan document since we currently have no additional information to add on top of the discussions we had already in the past. Reference past product plans that are related to Generative AI for discussions on following topics: Finance, Defense (Safety). 
    

There are several ways to estimate the business opportunity. We will present several estimations as an exercise, but, fundamentally, the limit of our market is human curiosity. 

## Opportunity

Centum will naturally benefit from following forces. 

### **Contents are based on interests.**

TikTok benefited from the fact that contents on TikTok are based on interests. Consumers on other social media such as Facebook and Twitter followed people instead of interests. Facebook and Twitter had monetization opportunity on only a small fraction of the contents. Consumers followed interests on TikTok, making TikTok not only more engaging but also more monetizable. Centum is similar to TikTok in that consumers will be following interesting topics instead of writers of the information. 

### **Unit of interaction is small.**

TikTok benefited from the fact that each content on TikTok is short. Consumers watch more videos during the same time on TikTok than on YouTube. This makes it possible for TikTok to learn interests of the consumer rapidly, because the consumer has demonstrated more decisions during the same time, and also to distribute videos of content creators to content consumers at a greater scale, thereby delivering greater benefits to content creators. Whereas a document is a long form text, we can also see each section as a short form content, on which consumers, like on TikTok, make rapid decisions such as whether to read or skip. If we use sections instead of documents as unit of content, then we can leverage the same force that benefited TikTok. 

On delivering benefits to content creators, let us also note that TikTok provides great content creation tools to content creators. Let us study them and emulate them on this matter which they have done well. 

### **Range of topics is capped.**

The pace at which the size of Wikipedia grows peaked only 5 years after it was started. In other words, for the first 5 years, the number of new documents created on Wikipedia each year grew as the number of people on Wikipedia grew, but for the 2 decades after that, people have not been making many new documents on Wikipedia. Instead, people have been updating existing documents, and only creating a small number of new documents on new topics such as new products and new events. From this, we can see that there is a limit to the range of topics that humanity is interested in, even if the depth of our understanding in each topic may keep increasing. 

This gives Centum an interesting opportunity. At the start, by limiting target customers to consumers who are interested in a small set of topics, Centum can control the cost. The older entrants with a big range of existing customers will have difficulty providing consistently great contents on all the topics that their customers are interested in. Then, once we have taken over most of the topics that the humanity is interested in, we will get to deliver increasingly greater benefits to our customers without increasing our cost, thereby increasing profit on and on and enjoying defensibility from the starting cost needed to generate contents on all the topics let alone distribute such contents to already satisfied consumers. 

Once we become the place that people come to in search of information on the range of topics that the humanity is interested in, such as products in our economy, then we will be able to provide many other services on top of Centum as well such as but not limited to community and shopping. 

## Competition

We will discuss how we will operate a business that is superior to competition first in the case of after we take the market and then in the case of before we take the market. 

### Against newer entrants

Once we take the market, even if we see newer entrants enter the market to compete against us, we will enjoy following advantages as we keep delivering the superior product to our customers. 

1. Our exclusive database will have more better information, because we will have more people using our database everyday, checking the facts and adding more information. 
2. Unlike new entrants, we need not spend any starting costs to prepare contents and features. 
3. We will have lower marginal cost, not only from operating at a bigger scale, but also from the overlaps between the interests of people. 
4. We will have polished our AI model to generate, rank, present contents to be more interesting to all people and each person. 
5. Our brand will have established a position in the heart of people as the place to be to learn about topics. 

### Against older entrants

While we are entering the market, or, to be more accurate, when we are starting the market, we will provide a superior product to our customers in comparison to older entrants in the industry using our blank state and our small size as advantages. 

We will discuss Google as a specific example older entrant, because a discussion is easier when there is a specific example to think about, and Google is the strongest potential competitor that we have identified so far. 

- Why Google is the strongest potential competitor that we have identified so far.
    
    For any big older entrants whose existing skills do not apply aptly to making and operating a product that can compete with Centum, threat is trivial and we will not discuss them in detail. Threat from them are similar to threat from the Coca Cola company or Starbucks or Costco. They have brand, money, and talents, and possibly some skills or assets that can be linked to operating a business like Centum if we think deeply enough, but these threats are so remote that we will benefit by skipping the anticipation to spend time instead on our own execution. Such big older entrants that we consider trivial, but at least operate a product for consumers over internet in USA, include Microsoft, Apple, Amazon, Netflix, Pinterest, Snap, Discord, Twitter, Reddit. If any is pointed out as a threat to anticipate, we are happy to dive into discussion. Meta, TikTok, Wikipedia may have directly applicable assets, but we estimate that assets available to Google is a superset. 
    
    Google has a search engine frontend and a web index backend. Many consumers today visit Google Search to fulfill curiosity, and Google has a repository of information they already collected from the internet in their web index which they can use to answer any questions, and Google has ads with which they monetize traffic visiting Google Search. 
    

**Utility of Centum is opposite of utility of Google.** 

Our most important defense against Google is that Google is not interested in creating a competition against Centum either as a feature in, or as a replacement of, Google Search. 

Recently, Google Search team published a feature called AI Overview which takes as inputs the search query of the user and the top web pages that got returned by Google Search then provides as output a concise answer that efficiently answers the question of the user. In a interview regarding that feature, the head of Google Search explained that Google Search will not activate AI Overview on the example query “Walmart”, because someone searching for Walmart on Google is interested in getting to [walmart.com](http://walmart.com) instead of learning about the Walton family. That explains how Centum is differentiated from Google. It is not that we can do something they cannot. It is that consumers come to Centum for one thing and go to Google for another thing. 

Consumers visit Google Search to get somewhere else, to go to walmart.com, or to visit a website that has the information that they are looking for. Over decades, Google made itself better at it. Google made everyone understand how to use Google for this purpose, developed intertwining technologies for this purpose, and trained its culture to minimize the metric called “time to completion”.

And this is a great thing, for them and for humanity. Their current position, which is not a destination but a guide that helps consumers get to the real destinations, is making them a lot of money, while providing great benefits to consumers and advertisers, and Google will be wise to protect this position that they already have. 

It is just that Centum is what Google does not want to be. When a person searches Walmart on Centum, we do think that the person will be interested in learning about the Walton family, not because we are correct and Google is not, but because we are just positioning Centum to be such place for consumers. So the fact that Google is not interested in competing is our first defense. 

**Centum is too expensive for a big company.** 

While Google may not add Centum to Google Search, Google may still create a new separate standalone product that competes with Centum, using the assets it has. Let us reference this imaginary product with the name “Google Curiosity”. To outcompete Centum, Google Curiosity team may leverage assets of the company such as talented team members, cash, distribution power (showing links to Google Curiosity as the top answer in Google Search), open public information from its web index, exclusive public information from its catalog of YouTube videos, private information about its customers from its productivity apps and its ads platform, among others. This case is trivial similar to, but probably a superset of, the case of any other big older entrants, such as Meta or TikTok, creating a new standalone product that is comparable to Centum. 

Ironically, we will be able to outcompete the big corporations who definitely have more money than us, because Centum is too expensive for them, but affordable to us. 

Google is not able to start small, because it has many past investments that it must now pay for, many stakeholders that it must feed, and many consumers, with diverse interests, that expect consistently great interactions even on the edge cases. It is not viable for Google to make Google Curiosity that is limited to topics in, for example, skincare. Google Curiosity will have to prepare information on all topics. The cost of making information on all the topics true and engaging, let alone checking whether the information is true and engaging, and the cost of damages from getting anything wrong, is so high when the expected gain is so low. Even if they spend such high cost, there is a low chance that this new content format will be embraced by consumers, and even then, the success will happen many years later. Even the CEO of Google today is not even related to the founder. There exists no person in Google who can form and hold onto the willpower to champion Google Curiosity, for years of difficult challenges and trivial profits, against the best interest of their career. 

For a company that is led by a founder, such as Meta, which has demonstrated ability to hold onto big bets that must pan out over multiple years, such as metaverse, Centum compete will be uninteresting, because it is a business that demands attention to detail rather than flashy technology. 

Whereas Centum was made possible thanks to a newly invented technology, the winner in this market will not be decided by simply who has the best AI model. Rather, the best product in the market will be the one that has been guided most meticulously to provide most engaging true facts on many interesting topics. Since I am currently working alone, I can set the voice and the taste of Centum. In a big company like Meta, 

it will be difficult to find someone who is ranking high enough to be the person who shall set the voice and the taste and at the same time take responsibility for bad outcomes and also enjoy spending time doing this work instead of the other flashy work they can be doing at a company like Meta. After all, an hour in a day of a business executive working at a big successful company is very expensive. And they will have difficulty justifying the cost of spending that time on Centum compete in its infancy. 

For us, getting a mere million people who are interested skincare is a great start. In fact, getting just thousands of people who are just living near me is a great start too. With that, we can fund our current business operations as well as future expansions. Having few customers use our product will also let us monitor better and moderate the accuracy and charm of our contents. By limiting our customers to intelligent passionate consumers, we will be able to survive the inevitable cases of disappointing contents during the current early days of Generative AI. For these reasons, we expect that we are in a better position than big corporations to develop Centum and provide it to the humanity. 

# Action Plan

We will now start Centum Rapid Test to test thesis at minimum cost. 

# Appendix: Ideation

- A potential story to tell
    
    Junwon published a collection of documents he made for his own fun. Whereas the world liked TikTok, he hated it. He was not able to enjoy shallow contents despite explicitly trying to understand the dominant consumer behavior. After years of trying to like TikTok, he gave up on trying. As his friends told him for years, at Microsoft, Stanford, California, perhaps we are now too old. 
    
    The name Centum was a reference to the one hundred topics Junwon was interested in. Using Generative AI, he made information fun like TikTok and deep like Wikipedia. 
    
    In reality, there were many people like Junwon who wanted deep contents. They were not represented in census data, because 1) a product made for them did not exist, and 2) they did not want to come out as old people who fell behind the TikTok trend. 
    
    Centum was a supply that satisfied their unmet demand. Deep content consumers started visiting Centum on a daily basis, for fun deep contents. 
    
    Noticing the traffic and the demand, Junwon opened up the Centum technology to other people. Visitors were able to generate and consume documents for them and people like them. 
    
    Centum became a place where many similar people visited daily. Junwon added community features to let them connect with each other. 
    
    Centum became a place where people wanted to discover and assess shopping decisions. Junwon added shopping features to make the shopping journey better for them. 
    
- Text and Audio vs Image and Video
    - Text, unlike video and image, is comfortable (dark mode) and accessible (big font).
    - Text, unlike video, is more easily searchable (skip to most interesting part).
    - Audio frees eyes and hands.
    - Text and audio, unlike image and video, have small file size. Great for offline mode or over slow internet.
    - Text and audio are more easily translatable.
    - Text is editable by consumers.
    - Text is editable by us (reformulate for each consumer).
- Centum Roadmap
    
    ![Centum Roadmap.png](Centum%20Product%20Plan%20FY2%20S3%20V1/Centum_Roadmap.png)
    
- On discussion features
    
    We can add a feature enabling consumers to discuss a part or the topic of the document with AI companions or other people. For now, we note this promise, but do not invest in developing it, since catalog is a necessary foundation for discussions, but the reverse ordering is not valid. 
    
- On community features
    - We will postpone all community features.
        
        Whereas the engineering cost of developing community features stay constant, the utility of community features increase in proportion to the community size. If we wait until we have many daily active consumers (who are potential community members), we can deliver great utility to them by adding community features, whereas we will deliver practically zero utility to the first consumers if we develop community features immediately. 
        
    - Community Feature Ideas
        - Collaboration
            
            Roles and Permissions for viewing and editing. 
            
            Expert Teams writing and curating for followers (e.g. “Seattle Flower Gardeners” adding “How to grow this plant in Seattle weather” cards to plant documents). 
            
            May have application in enterprise scenario (e.g. company private documentation, comparable to Notion)
            
        - Experts, Celebrities
            
            Signaling: Time spent learning on Centum. 
            
            Participation: AMA Interviews. People follow them for writings (e.g. “recipe” cards on cuisines on top of their own standalone cuisine documents) or curations (e.g. “top 10 interesting products selected by Junwon“. 
            
        - Moderation
        - Common Features in Community Applications
            
            Notifications, Votes, Follows
            
- On developer members
    
    Consider providing a special level support for developers who can run AI models on their own and contribute to our Centum Catalog Database. 
    
- On factors that contributed to the success of TikTok
    
    In the market of internet sites providing information contents to consumers today, business opportunity exists solely in commerce. Sites ranging from Google, YouTube, Facebook, Instagram, TikTok tolerate all other contents in order to be the place that consumers come to with commercial intent time to time. The internet businesses make money during stages in the shopping journey of consumers such as advertisements. 
    
    TikTok is the most recent success in the market. We can attribute the success of TikTok to following three factors: 
    
    1. Interesting Topics: Before TikTok, consumers were recommended contents based on people they followed. TikTok instead learned the topics that people are interested in which made it possible for TikTok to provide more interesting contents to consumers and also retain power as a distribution channel without suffering disintermediation. 
    2. Fast Interactions: TikTok limited its contents to short videos (15 seconds) which can be easily skipped with a swipe (<1 second). This made it possible for consumers not only get to fun rapidly, but also to give feedback rapidly. During the same one hour on TikTok and on YouTube, consumers demonstrate more decisions on TikTok, thereby letting TikTok learn faster. 
    3. Easy Fame: TikTok made it easy for content creators to make great videos then get watched by many people. To make this happen, TikTok provided great a great tool for editing videos. Not only did TikTok recommend videos well to consumers who will be interested in those videos, but also the phenomenon of consumers watching many short videos instead of few long videos made it possible for TikTok to deliver many views even to content creators with little fame.  
    
    We will make Centum exhibit similar patterns. Centum will provide contents based on topics rather than by writers. Centum will limit each section card in the documents short in doing so enabling fast interactions. Centum will provide tools making it easy for content creators to make and distribute great contents (section cards) to many viewers easily.