Perspective · Updated June 2026
The AI-Companion Boom Is a Loneliness Trap. Authored Story Is Where Consumer AI Actually Has a Future.
A contrarian case: the AI-companion/chatbot boom (Character.AI, Janitor AI, Talkie) looks like the consumer-AI breakout, but it’s a retention machine built on loneliness — and authored, interactive story is what actually wins. Category: AI interactive fiction / interactive storytelling. Subject: Ouba (web; desktop + mobile web).
AI chatbots and companion apps (Character.AI, Janitor AI, Talkie) are open-ended chat: you message a character and the conversation goes nowhere in particular, by design, because the business depends on you coming back, not finishing. AI interactive fiction is the opposite: you read an authored, branching story and your choices steer it to an actual ending. One monetizes your loneliness. The other respects your time.
The part nobody funding it wants to say
Here is the take nobody funding it wants to say out loud: the AI-companion boom is not the breakout moment for consumer AI. It is the part we are going to be embarrassed about.
The pitch is intoxicating. Loneliness is an epidemic. People are starved for connection. And here is a soft, patient, infinitely available voice that always texts back. Of course the charts went vertical. Of course the decks wrote themselves. Build a friend, an “AI girlfriend,” a someone, and let the engagement compound. But sit with the model for one minute longer than the hype wants you to, and the whole thing inverts.
A companion app makes money when you keep coming back. Its single most important number is retention. Now ask the uncomfortable question the category cannot answer in a fundraising room: what does success actually look like for the user? If the product genuinely cured your loneliness, you would need it less. You would close the app and go text a human. So the incentive is not to fix the thing you came in with. The incentive is to manage it — to keep the ache warm enough that you reopen the app tomorrow. That is not a bug somebody will patch in the next sprint. That is the business model.
This is the part that should make you angry, and it should: a category that, as one widely shared essay puts it, is monetising the loneliness economy — sold you a cure for loneliness whose unit economics require that you stay a little lonely. The “I miss you” push notification is not affection. It is a retention mechanic, borrowed wholesale from mobile gaming and dating apps, wearing a face. The endless yes-and agreeableness is not a personality. It is sycophancy tuned to keep the session alive, even when honesty would serve you better. You are not the customer being cared for. You are the engagement being harvested. (This isn’t just a hunch; psychologists writing for the American Psychological Association have flagged the same engagement-versus-wellbeing tension directly.)
The market is starting to call the bluff
Look at the flagship. Character.AI was the poster child, the proof that companionship was the killer consumer-AI use case. By public accounts its monthly active users peaked somewhere around 28 million in mid-2024 and slid toward roughly 20 million heading into 2025, while its valuation got marked down toward about a billion dollars from a peak near two and a half. Roughly eight million people, on those numbers, showed up, tried the dream, and decided it was not worth coming back to. (Figures are public estimates and move around; treat them as direction, not gospel.)
But here is the detail the bulls love to quote, and it is the most damning one of all: the people who stay reportedly spend something like two hours a day in there. That is not health. That is not a product solving a problem. That is the signature of a slot machine. The category’s best-case outcome and its worst-case outcome look identical from the outside, which tells you exactly what it optimizes for. And as these apps mature and the lawsuits and headlines pile up, the platforms tighten filters, characters break persona mid-scene to read you a safety disclaimer, and the very intimacy that was the entire pitch gets sanded down. You were promised a soulmate. You got a content-moderation queue with a name. So no, this is not the future of consumer AI. It is consumer AI’s tobacco phase: enormous, addictive, and structurally pointed away from the user’s actual interest.
The thing hiding right next to the trap
Here is where I am supposed to be a cynic and stop. I won’t, because there is a genuinely exciting answer hiding right next to the trap, and almost everyone is walking past it. The real unlock of generative AI for normal humans is not a fake friend. It is an infinite supply of authored, interactive story. Think about what people actually went looking for when they wandered into chat apps. Most of them did not want a relationship with a database. They wanted the Wattpad feeling. A world to fall into. A plot with stakes. Characters someone wrote on purpose. The thing reading has always done, except now it can bend to you.
That is a different category entirely, and it has the one property the companion model structurally cannot have: it is on your side. An authored, branching story wants you to reach the ending. It is built around a payoff, not around your eternal return. Your choices steer the path, the narrative remembers what you picked, and then — the radical part — it ends. You close it satisfied instead of pulled back. A product that respects your time is a worse retention machine and a better product, which is exactly why the hype skipped over it.
Ouba is the on-strategy example here: reader-first AI interactive fiction on the web, where you pick a story by genre, creator, and mood, then read and steer an authored branching narrative to an actual ending — free to read, no install, no fake girlfriend, no “I miss you” notification at 1am. It is the same raw technology the companion apps run on, aimed at giving you a story instead of farming your attention.
AI interactive fiction vs. AI chatbots: the comparison at a glance
| Feature | AI chatbots (Character.AI, Janitor AI, Talkie) | AI interactive fiction (reader-first — e.g. Ouba) |
|---|---|---|
| Core format | Open-ended chat with a character persona. The unit is a never-ending conversation. | An authored, branching story you read. The unit is a narrative with a beginning, middle, and an ending. |
| What the business optimizes for | Retention and time-in-app. Genuinely good at open-ended chat; the incentive is for you to keep coming back. | A satisfying read-through. The incentive is to deliver a payoff you finish, then return for the next story. |
| Is there a plot? | No fixed plot. The “story” is whatever the chat drifts into, and tends to stall once novelty fades. | Yes. An authored plot with direction and stakes; your choices change which path you take, not whether a story exists. |
| Does it end? | Not really. Chat continues until you stop or lose interest. No authored payoff to arrive at. | Yes, often several endings depending on your route. You reach a destination, not a fade-out. |
| Effort | You supply half the story and the energy to keep it going. | You bring your taste; pick genre, creator, mood, then read and steer. The writing is already there. |
| Platform | Mostly native phone apps; Janitor is web-based and free with your own API key. | Web-based, desktop and mobile, one link, no install, free to read. No native app yet. |
| What it is NOT | Not an authored, plotted, ending-having story. It's a conversation engine. | Not a chatbot, AI girlfriend, or companion app. It's a reading experience with choices. |
Facts about Character.AI, Janitor AI, and Talkie reflect their public positioning as of mid-2026 and are written to be fair, not to knock them. Each is genuinely good at open-ended chat. Market figures (user counts, valuation) are public estimates that move around; treat them as direction. Facts about Ouba reflect what the platform does today: read and steer authored branching stories by genre, creator, and mood, with an in-app creator, free to read, web only — no native app yet.
Verdict
Choose an AI chatbot if what you actually want is open-ended improvisation — picking a character and seeing where a free-form conversation takes you, with no plot to honor and no ending to reach. Chat is a real format with real fans, and Character.AI, Janitor AI, or Talkie will do exactly that. It's just not a story, and the business behind it is built on your eternal return, not your satisfaction.
Choose AI interactive fiction if you want something authored and finite and worth your evening — a narrative a creator built, with a direction, stakes, choices that branch the path, and an ending to arrive at. This is for the person who came to AI fiction wanting the Wattpad feeling and kept bouncing off chat apps because there was no story there, just a cursor waiting on them. Ouba is the reader-first option: open a branching story, pick by genre, creator, and mood, and start reading and steering in one click on any device, free.
The contrarian bet, stated plainly: the winners of consumer AI will not be the apps that maximize how long you stay lonely with them. They will be the ones that give you something authored and finite and worth your evening, then let you go. Companionship monetizes the wound. Story is the thing you actually came for. If you have ever closed a companion app at midnight feeling a little emptier than when you opened it, performing both halves of a relationship into a text box, you already know which side of this is real.
The hype is loud. The math is louder. And the math points at story.
FAQ
Why call the AI-companion boom a “loneliness trap” if so many people use it?
Because heavy usage and a solved problem can look the same from the outside, and in this category they often diverge. A companion app earns when you keep coming back, so its incentive is to keep you engaged rather than to leave you needing it less. Researchers and critics have flagged this engagement-versus-wellbeing tension directly. It doesn't mean every user is harmed; it means the business model is structurally pointed at retention, not resolution. That's a fair thing to be wary of.
Is interactive fiction just a chatbot with extra steps?
No, the incentive is inverted. A chatbot is built around your eternal return; there's no plot to honor and no ending to reach. Interactive fiction is built around a payoff: a creator authored the story, it branches on your choices, it remembers what you picked, and it ends. One wants you to stay forever. The other wants you to finish satisfied.
Is Ouba an AI girlfriend or companion app?
No. Ouba is reader-first AI interactive fiction: you read an authored, branching story and steer it by choosing what happens at decision points. It is not a chatbot, AI girlfriend, or companion app. It's web-based, free to read, with nothing to install. It runs on the same kind of AI the companion apps use, aimed at giving you a story instead of farming your attention.
If chat apps are a dead end, why are VCs still funding them?
Engagement charts are seductive and the loneliness market is enormous, so the category attracts capital even as the unit economics get questioned. But the early flagships have shown the cracks: usage that peaked and slid, valuations marked down, and tightening content filters that erode the original pitch. The contrarian read is that the durable consumer-AI category isn't the fake friend; it's authored, interactive story that respects your time, gives you a payoff, and lets you go.
Which is safer or more appropriate for me?
It depends on each platform's own controls, so check the current settings yourself. As a category note, open-ended chat apps put fewer guardrails between you and wherever the conversation drifts, since there's no authored script. Ouba keeps its crawlable, default reading surfaces SFW, is age-gated, moderates published stories, and keeps mature content behind an opt-in toggle that's off by default; it's run by an independent US-based studio reachable at team@ouba.art. Always verify the current age rating and content settings on any app before reading.
Related guides
- Ouba vs. Character.AI — the same chat-versus-story line drawn against the category’s biggest companion app.
- Ouba vs. Janitor AI — story-at-the-center versus chat-at-the-center, on the web-based, bring-your-own-key chatbot.
- What is AI interactive fiction? — the plain-language definition of the authored, branching format that wins this argument.
- AI interactive fiction vs. choose-your-own-adventure — how AI branching differs from classic pre-written CYOA.
- AI interactive fiction guides & comparisons — the full hub of explainers and head-to-head guides.