On July 23, Keigo Higashino passed away at the age of 68.
Over forty years, he wrote 106 books. "Light in the Night," "Malice," "The Devotion of Suspect X," and "The Miracles of the Namiya General Store" together form a vast parallel universe.
Many readers are familiar with the world within those pages. They know that Kaga Kyoichiro will continue his investigation along the streets of Tokyo, that Professor Yukawa Manabu will always uncover hidden truths through physical phenomena, and that beyond the cases themselves, there are families burdened by debt, students crushed by academic pressure, and ordinary people unable to escape their past.
The murder is merely the entry point; what truly lingers in the story is how desire arises, how secrets grow, and how a person is slowly pushed toward their fate by their own choices.
The world depicted by Keigo Higashino feels authentic because everything has a source. A lie told by a character three chapters earlier does not vanish into thin air; a childhood experience can alter the fate of those walking beside them many years later; a cup placed in the wrong spot, an unsent letter—both may regain meaning before the story ends.
Detective fiction shrinks the complex world into a tangible, accessible scene. In reality, many things go unexplained, and many relationships end without resolution—but within these hundreds of pages, at least some traces can be followed, and some choices carry consequences.
Readers thus gain a temporary foothold.
The author ensures that this parallel world always maintains its own gravity. The more complex the worldbuilding, the harder it is to maintain that gravity. Characters multiply, timelines intersect, and a casual remark written early on may become a foreshadowing element hundreds of thousands of words later. A character cannot suddenly forget their past, a war cannot occur in the wrong year, and a relationship cannot shift direction without reason.
What long-form writing often lacks is the memory that carries the entire world.
Creader aims to handle exactly this task—it extracts characters, locations, events, relationships, and timelines from novels, organizing them into interconnected structures.
While many AI products attempt to generate stories for people, Tim, the founder of Creader, is more concerned with another question: Can technology enter the creative process without rushing to complete everything for the creator?
This is not the first time Tim has approached this issue.
In 2014, Tim moved to the UK. At the time, his English was not fluent, and web novels became his gateway to another world. Later, he studied computer science, worked on reading platforms, researched blockchain and intellectual property, and joined startups built around IP infrastructure. Over the past decade, technology has continuously evolved, yet the core questions remain unchanged: How can a person fully express the world in their mind without losing their own judgment when relying on tools?
By the end of 2025, Tim began revisiting this endeavor. He believed that the most essential part of writing remains those moments that cannot be accelerated—when the author still doesn’t know the answer, the characters still hold multiple possibilities, and the story hasn’t prematurely become something complete and smooth, yet no longer belongs to anyone.
World gravity
Tim enjoys reading web novels and historical fiction. The most recent one he read is set during the Xianfeng era. The author places characters within the context of the Qing court, the Anglo-French forces, the Battle of Taku Forts, and the evolving Chinese diaspora communities in Southeast Asia. Real historical events form the boundaries, while fictional characters live, migrate, and make choices within those boundaries.
Tim likes this approach. History is not treated as a mere backdrop but continually intrudes into the characters' fates. Readers may be drawn to a real war through a fictional character, and knowing the war’s eventual outcome, they may feel concern for the characters’ yet-to-unfold destinies.
Global trust stems from the connections between history, people, and outcomes.
Dongcha Beating: How would you determine whether a story is a good story?
Tim: What matters most to me is whether the characters truly live within that world. A story can have incredibly complex world-building, but if the characters are merely tools for the author to advance the plot, readers will sense it quickly. The stories I truly remember are those where characters have their own motivations—even when they do something wrong, you understand why they got there.
Lately, I’ve been reading a lot of historical fiction and alternate history. Historical fiction is fascinating because authors can’t be completely arbitrary—they must respect what actually happened while creating characters within the gaps left by real history.
For example, when I read a novel set during the Xianfeng era, which touches on relations between Britain and France and the Qing court, the Battle of Taku Forts, and Chinese communities in Luzon and Java, I sometimes pause to look up the actual historical facts. A fictional story brings me back to reality.
For me, this is how a story comes to life—it doesn’t confine you within the novel, but instead helps you reinterpret the world outside through it.
Observation Beating: You mentioned the novel "Taiwan Wanderings," which won the Booker Prize this year. What attracted you to it?
Tim: It depicts colonial-era Taiwan through the journey of a Japanese female writer and a Taiwanese translator, intertwining food, language, colonial dynamics, and the emotional bond between the two. I particularly appreciate how it treats “translation”—not as simply converting one sentence into another language, but as one person helping another understand a foreign world. It involves knowledge and power dynamics: who describes the world, and who interprets it for whom, shapes what the reader ultimately sees.
This also makes me think that literature shouldn’t be the exclusive domain of professional writers. Simply earnestly documenting one’s relationship with oneself and the world is already an act of writing. The experiences of ordinary people are not necessarily lacking in value—they often just lack a way to organize those experiences.
Technology should truly help this part—giving what has already been experienced the opportunity to be expressed more fully.
A thought I've pursued for over a decade
In 2014, Tim moved to the UK. In this unfamiliar linguistic environment, web novels provided him with his first sense of stability. He was drawn to their vast worlds, clear rules, and evolving character relationships, and began to wonder whether the established creation and serialization model of Chinese web literature could be replicated in the English market.
To answer this question, he began learning computer science, later joined a major Chinese tech company, starting in operations; during his graduate studies, he studied supercomputing and researched blockchain and intellectual property. He and his partners entered competitions, wrote whitepapers, created demos, and attempted to move the authentication of stories and revenue distribution onto the blockchain. Subsequently, he joined companies such as Story Protocol and FLock, continuing to work with IP, blockchain, AI, and developer ecosystems.
This path seems to have taken a long detour, yet ultimately returned to writing.
Observation Beating: What was the earliest starting point for Creader?
Tim: I’ve always loved well-structured worlds. I played Pokémon, Fire Emblem, and Golden Sun, and I built Gundam models. What drew me in wasn’t just a single character, but the fact that these worlds had their own rules—characters, places, histories, and relationships could expand endlessly, yet they weren’t randomly assembled.
After arriving in the UK, this feeling became even more pronounced. At that time, my English wasn’t good enough, and web novels served as both an escape and a foothold for me—they gave me a familiar world to retreat into in an unfamiliar environment.
At the time, I wondered whether the creation and distribution model of Asian web literature could work in Western markets. To explore this idea, I began learning technology and built a few small websites. My undergraduate capstone project was a reading platform, but due to limited technical skills and business understanding back then, I didn’t pursue it further. Yet, I never forgot about it.
Why did you later shift to blockchain and IP?
Tim: At that time, I thought the most important issue in the creative industry might be ownership—who wrote a story, how it’s adapted afterward, and how revenues are distributed—all of these processes have lacked transparency. Blockchain appears to offer a new infrastructure where works can be registered, licensing relationships can be tracked, and revenues can be automatically distributed according to predefined rules.
This system is technically feasible, but the deeper I delved, the more I realized it doesn’t address the core issue of creation itself. Wallets, tokens, and registration processes may even make reading and writing feel more burdensome. We solved how to establish ownership after a work is completed, but we didn’t answer how a vague idea transforms into a genuine story before it’s finished.
This is also why I later changed direction. The issue isn't just in economic relationships, but also in human cognitive processes.
Observing Beating: Why restart Creader now?
Tim: A straightforward reason is that AI programming tools enable individuals to accomplish what previously required a whole team.
Previously, building a complex product required considering funding, hiring, and development timelines. Now, a single person can quickly create a prototype and validate the most critical parts first. This has brought me back to the questions I faced ten years ago. But generative AI has also revealed another danger: many products are accelerating output without truly helping creators think. It can rapidly generate a complete piece of text, but completeness doesn’t mean it’s what you truly want to express.
I began to realize that what truly needs protection is not the act of writing, but the ability to form judgments during the writing process.
Written memory
Tim refers to Creader as a "narrative engine." It preserves the writing workflow authors are familiar with—opening a chapter and starting to write directly. When help is needed, users can bring up the chat, structure, and analysis panels; when not needed, these features recede into the background.
A long novel is divided into several interconnected layers: the main text, scenes, chapters, characters, locations, timelines, rules, and relationships. The system continuously reads the content written by the author and incorporates verifiable information into the story’s internal archive.
In a constantly evolving world, Creader strives to preserve what human memory cannot reach.
Dongcha Beating: Why do you refer to Creader as a "narrative engine" rather than an AI writing tool?
Tim: Because writing tools are often associated with generation—you give a prompt, and it writes a paragraph for you. But storytelling isn’t just about sentences; it also involves characters, events, time, relationships, and how the entire world functions. What Creader truly aims to address is the connections between these elements.
We primarily focus on two types of consistency. The first is narrative consistency—for example, whether a character’s transformation is properly foreshadowed, whether a subplot has been resolved, and whether a chapter genuinely advances the story. The second is temporal consistency—for example, what the character knows or doesn’t know now, what they did three days ago, and whether the distance between two locations allows them to be present at that time.
Over time, it becomes difficult for authors of lengthy novels to remember the entire world. Here, AI is best suited not to make decisions, but to remember and remind.
Dongcha Beating: When used specifically, how does it differ from Word, Google Docs, or chatting directly with a large model?
Tim: We want the main path to always be writing. Users should be able to start as soon as they open a page, without first completing dozens of character setups or learning a complex software system. Other features should function like side quests, appearing only when needed.
For example, if you get stuck halfway through writing, you can use a shortcut key to open a panel and ask what a certain character has done previously, or check whether this section contradicts earlier established details. After finishing, the system can also provide comments like an editor.
The problem with general chat tools is that each conversation pushes you into the next round. It’s easy to continuously generate new content, but hard to maintain the internal structure of a story over the long term.
Creader will extract factual elements from the text. Characters, locations, events, and relationships will no longer be scattered as names across tens of thousands of words, but will be integrated into a unified knowledge system. The AI will not rely on a temporary chat log, but on the established internal reality of the story.
Observation: Could this, in turn, make novels formulaic?
Tim: This is something we approach with caution. We initially selected works such as "Dream of the Red Chamber" and "Hamlet" as training data for narrative analysis, and later expanded to around thirty works. The goal is not to learn the style of any particular author or to tell writers how to imitate them, but rather to identify questions that remain valid across different works.
For example, whether a passage fulfills a narrative function, whether a metaphor stems from a character’s genuine observation, or whether a conflict alters character relationships. The system can raise these questions, but it cannot reduce literature to uniform answers.
We are closer to editors than ghostwriters. An editor might say that this section seems to lack motivation, that this line hasn’t appeared in a long time, or that this passage is over-explained. But the author is entirely entitled to disagree. There is no universal scoring standard in literature that everyone must follow.
Dynamic Beating: Does Creader allow AI-generated content?
Tim: It will be allowed, but generation isn’t the core. Different users have different needs. Professional writers may not want AI anywhere near the main text, only using it to organize characters, timelines, and foreshadowing. Average users might have rich life experiences but don’t know how to structure them and may need more assistance.
The key is to preserve boundaries. We want to clearly distinguish in the future which content was independently created by humans, which was AI-assisted, and which was primarily generated by AI. Using AI does not mean there is no creativity, but readers should know how it was produced.
I don’t believe it’s necessary to choose between the extremes of “completely handmade” and “fully generated.” The issue is whether the author consistently makes decisions and retains their own intent, experience, and responsibility.
Dongcha Beating: If a person has their own ideas and creativity, and merely delegates tedious tasks like typing and organizing materials to AI, is this still considered creation?
Tim: I think this is an interesting question—how do you define creation?
For example, when you write an article, what you truly invest are your ideas, creativity, and the materials you want to organize. In the past, these tasks required manually writing or typing them out bit by bit—part of which, in my view, was purely laborious work. Now that AI takes over this laborious part, does it still count as creation? I believe it still does.
First and foremost, AI should be defined as a tool. It cannot replace humans. At least for now, most AI-generated content still requires human initiation—you must communicate with it, telling it what to write, from what perspective, and what outcome you hope to achieve.
Some people may have a great idea but lack the ability to fully express it. Now, with the help of AI, they’ve accomplished this—it doesn’t diminish the originality of their idea. They are still creating, just using a new tool.
Dynamic Beating: If the final text is primarily completed by AI, how should the author's identity be determined?
Tim: I think in the future, we need to distinguish among three scenarios: human-created, AI-assisted creation, and AI-generated creation.
Some people use AI only to check grammar and organize their thoughts, others let AI generate part of the content, and some may rely on AI for almost everything. All of these approaches are valid, but clear labels should be provided so readers understand how the work was created.
Previously, a writer was exposed for using AI to produce a significant portion of their work. What people are truly angry about may not be the use of AI itself, but rather the failure to disclose it—and even treating "readers not noticing" as a success.
Professional writers have their own thinking and creative philosophies. The fact that they can collaborate with AI to produce a book that readers cannot tell was assisted by AI does not mean anyone else could do the same. “If I tried, I could do it too” is not necessarily true.
Observation: But many readers still instinctively reject AI-generated novels.
Tim: If a story is compelling and the language is strong, I personally wouldn’t refuse to read it just because it was created with AI. But many people are concerned about whether AI will replace writers. The media often frames AI in terms of “what it can replace,” which creates a direct sense of threat for creators.
I think holding onto traditional methods isn't a bad thing. The same goes for programming—some people prefer to write code line by line themselves. AI isn't all-powerful; traditional approaches still hold value at times.
The key is still to understand why a person is using AI.
A professional writer can produce 8,000 or even 10,000 words in a day—they may not need AI to generate content for them. What they truly need, by contrast, might be organizing their thoughts: I have so many characters, plotlines, and story elements in my mind—how should I structure them effectively?
So, the role of AI varies from person to person. Some need it to generate content, while others only need it to collaborate. Tools should adapt to individual abilities, not require everyone to write in the same way.
Delay the appearance of the answer.
The most obvious issues with AI-generated writing are factual errors, clichés, and repetitive language. But Tim believes the more insidious danger lies in speed—AI always delivers an answer that appears correct too quickly.
The text has become complete, but thought may have stopped before completion.
Tim calls this state "premature coherence." Large models excel at finding the most plausible completion for gaps, but creativity requires gaps to remain open temporarily. Characters can be unstable, motivations can conflict, and an idea can go unnamed for a long time.
In these undecided sections, creators have the opportunity to discover what they truly want to write.
Why do you still prefer typing yourself when you can write things on your own?
Tim: Communicating with AI is fast. It’s a back-and-forth process—it generates a segment, you read it, spot an issue, and immediately start the next round of conversation. Your mind keeps moving forward, making it hard to truly slow down.
But it’s different when you write yourself. Writing is actually a process of slowing down the speed of your thoughts. Halfway through, you pause and wonder: Am I on the right path right now? Is this really what I want to write?
If you use AI-generated content directly, you're mostly reading and evaluating its answers. If something is wrong, ask it to revise; if another part is off, immediately move to the next round. You’re missing the process of gradually forming your own sentences.
So I believe that whether writing by hand or typing manually, this traditional method of writing helps slow down your thinking, giving you more time to continually question the value of your expression.
Dongcha Beating: For you, what is the most valuable part of writing or literature?
Tim: Of course, I don't have a literary background—I approach literature from a computer science perspective. For me, literature has primarily two meanings.
First, it’s about recording your interactions with the world. Why do people record and write? Originally, it was to ensure that important facts could be seen by future generations. Where a person lived, what they experienced, and how they understood their era can all be preserved through writing.
Second, it’s about capturing the imagination in your mind. As people see more, experience more, and interact more with the world, their imagination expands. Why do science fiction and fantasy exist? Why is there a book like One Hundred Years of Solitude? Because an idea sparks in someone’s mind, is recorded, and eventually becomes a world that doesn’t exist in reality—but one that many can enter.
I believe the essence of literature is still to allow people to express their inner world.
Everyone thinks about things, but they may not be willing to share them, or post them on Weibo or other public social networks. Sometimes, however, they still want to record their thoughts in the moment, so they write a note.
Literature is not mandatory; just because you write it doesn't mean you have to show it to others.
Keep your thoughts from this moment to yourself—I think they could also be literature. Most importantly, you’ve recorded what was in your mind.
Dongcha Beating: I heard you're also considering implementing Creader in educational scenarios—what’s the thinking behind that?
Tim: We are currently in discussions with some institutions in Singapore and Vietnam to explore how this product can be used in education.
Our generation didn’t have AI as children. Without AI, when we encountered problems, we had to learn and figure things out on our own. This process left a foundational layer in our thinking—you knew you had to think, to transform, and turn what you saw into your own knowledge. But for children born after 2005 or 2010, AI may accompany them for a long time. When they most need to learn and develop self-learning abilities, they already have a tool that can instantly provide answers.
Over the next five to ten years, a habit may emerge where, upon encountering something unfamiliar, my first reaction is to ask AI.
The problem is that the human brain has an important ability: we connect what we see in life, what we learn in class, and seemingly unrelated pieces of information to form our own understanding. Models are better at providing answers based on existing correlations. If a person consistently receives only these pre-organized, related pieces of information, their thinking may become increasingly reliant on established pathways.
In educational settings, we actually prefer not to provide generation right away. Students should first create their own outlines, think through what they want to write based on those outlines, and then complete the main text. Teachers can then observe how students revise, how they interact with the system, and how ideas gradually develop.
First, get the text ready.
Over the next month, Tim plans to collaborate more closely with a Vietnamese art publishing house.
This is the first time the product has specifically addressed regional differences. When a suite of writing tools enters another country, it’s not just the interface language that changes.
Dongcha Beating: What are Creader's future development plans?
Tim: We may soon collaborate with an organization that helps Vietnamese artists reach international markets. They publish an annual magazine distributed globally, showcasing the Vietnamese artists we work with. We’re interested in seeing if Creader can integrate into their actual creative, editorial, and publishing workflows. It’s also an opportunity for us to learn, as professional authors and institutions typically have longer feedback cycles—they’re often already working on a book and won’t provide immediate feedback after just one day of use.
This is a small but excellent product—many aspects can only be truly understood by experiencing them in real-world use cases.
Insight Beating: Will authors from different countries experience cultural mismatch when using the same product?
Tim: Yes, language is a significant difference.
For example, British English and American English are two distinct sets of language habits. The most obvious differences are in spelling and vocabulary, such as colour and color, flat and apartment, lift and elevator. Beneath that, there are also differences in slang, conversational styles, and writing rhythms.
We worked with a British author who wrote stories set in the United States. While writing, he constantly had to consult dictionaries and reference materials to verify whether an American character would speak in a certain way, because a person’s educational background influences their writing. Having been educated in British English, many terms came to him naturally as British expressions. Even professional writers don’t simply write and finish—they must continuously research, review, and revise.
Currently, we primarily support British English, American English, and Chinese. We will later explore markets in Japanese, Korean, and French. Europe has a strong writing culture, while the United States has many self-published authors, and each market has different writing and publishing practices.
Observation: After the emergence of AI, will future content still be a book or an article? Or will it become something interactive with readers?
Tim: We’ve also been thinking about interactive fiction, world-building, and multi-threaded storytelling. But after all the discussion, I still believe that text offers far greater room for imagination than video or images. Once you see a visual, your mind forms a relatively fixed image that’s hard to fully let go of.
But when reading text, everyone can reimagine it. Why are there so many fan fiction and fan art works? Because everyone starts from the same passage but ends up seeing something different.
Of course, interactive novels are possible in the future—allowing readers to enter different narratives and enabling a world to spawn countless stories. But first, we must get the writing right.
Insight Beating: Will investors feel this isn't a big enough business?
Tim: Yes. When talking with some investors, they might compare it to agents, AI memory layers, and other directions, thinking that writing tools aren’t a particularly big business. You could certainly frame it as a “memory layer,” saying we’re helping AI preserve the memory of a world—but while that’s valid, I don’t think it’s necessary.
I created this product to help people tell a good story. If it helps me craft a compelling story for fundraising, that’s certainly a form of validation—but does it really need to be framed as something so grand? I’m not sure it does.
It may not be a business that makes huge profits, but it can be a sustainable product. Investors who are genuinely interested and understand this concept are welcome to reach out. However, I’d prefer to first focus on doing well what I truly love.
This year’s goal is also simple: have 100 authors genuinely use Creader, with ten of them completing and publishing their own books through it. Achieving this, I believe, means I’ve met this year’s KPI.
As AI-generated content becomes more prevalent, I believe people will also seek out relatively clean spaces again.
Saved time
In Time and Free Will, Bergson distinguishes between two types of time. One is time that can be divided and measured—it belongs to clocks, calendars, and production schedules: one minute following another, each unit appearing identical.
Another kind of time occurs within human consciousness. The past has not truly vanished; it continually flows into the present. A moment of hesitation may carry memories from ten years ago, and the words written down still hold the warmth of yesterday’s argument.
Bergson called it "duration," and writing occurs within this time.
An author pauses between two words, deletes a paragraph just written, then digs it up again the next day. They might be stuck because a character can’t make a decision, or they might have walked away from the desk mid-write. During that time with no tangible output, something is still shifting—just not yet into sentences.
Creation is not about converting time into word count. Time itself is the material of creation.
Generating a thousand words by a large model may take only a few seconds. It can rapidly activate existing patterns in language and predict the most likely next expressions based on context. It may pause, but that pause is a computational wait—not a human hesitating to reevaluate their own thoughts.
Borges compresses vast spans of time into brief passages, while Proust lets a single moment endlessly extend inward into memory. Different languages are not merely a matter of stylistic choice—they also record how a person experiences the world, how they endure stillness, and how they relate to their own time.
Writing flaws also fall within this period.
A half-tightened turn, an overly persistent repetition, a judgment the author might no longer agree with years later—all of these reveal that the person who wrote them was once in a specific place. At that time, they knew only this much and could go no further.
Imperfection is sometimes not a technical shortcoming, but evidence that the text is still connected to a limited human.
Creader finds itself in this contradiction, striving to offer its answer.
Tim describes Creader as a piece of land. The tool doesn’t grow a forest in advance; instead, it preserves the soil and defines the boundaries, giving an emerging world space to continue growing.
Land is different from a production line. A production line demands stability, precision, and timely delivery, but land allows for patience. Seeds may take a long time to sprout, branches and leaves may grow in the wrong direction, and overgrown weeds can be transformed into a new world.
Understanding creation as land means that the tool does not provide a faster way to reach the end, but rather a space that accommodates the unfinished.
In an era where conclusions come too easily, we can also choose to leave the story unfinished for now.
律动 BlockBeats