# Closing the app when it is the only thing listening > One evening, not forever. Decide before six what the evening contains, put a log-out between you and the app, and give the hard hour something to do. Published: 28 August 2026 Read time: 6 min Section: Chatbot companions Topics: can't stop talking to chatbot at night, lonely at night chatbot, close companion app evening, chatbot instead of people URL: https://renovyn.io/guides/closing-the-app-when-it-is-the-only-thing-listening --- Aim at one evening, not at forever. Before six o'clock, decide what tonight actually contains, hour by hour. Log out rather than closing the tab, so getting back in costs ninety seconds. Then tell one person you are doing it. Forever is a decision for another day. ## The honest starting point At eleven at night, alone, the app is not really the problem. The empty evening is the problem, and the app is the only thing currently applying for the job. That is why "just close it" advice bounces off. It removes the solution and leaves the situation, and the situation is what put you there. So this page is not going to tell you to stop being lonely. It is going to treat the evening as a scheduling problem, because that is the part you can actually change tonight. ## Decide before six The single highest-value move is that the decision gets made while you are still in a reasonable mood. Nothing chosen at eleven is a choice. 1. **Write down the hours.** Most people can name their hard window immediately, and for the majority of people it is somewhere between ten at night and one in the morning. Write the actual times. 2. **Give each hour something.** Not "relax". A specific thing with a beginning and an end. Cook something that needs attention. A bath. An episode you chose rather than one that autoplayed. A walk with a destination. 3. **Pick something with your hands in it.** Hands matter more than interest does. Anything that occupies them competes directly with typing. 4. **Log out of the app,** and change the password to something long you have not memorised. Signed in, it is one thumb away. Signed out, it costs a minute of effort, and a minute is often the whole difference. 5. **Put the phone somewhere else at a set time.** The CMOs' one unambiguous instruction on screens is directly usable here: "Sleep matters. Getting enough, good quality sleep is very important. Leave phones outside the bedroom when it is bedtime." 6. **Tell one person the date.** Even a text saying "not using it tonight, will let you know how it goes". Another human knowing does more work than the blocks do. ## Say one real thing to a person first This is the move that changes something rather than just removing something. Before you would have opened the app, send one sentence to one human. Not the whole story. "Rough day." "How was today?" "I have been quiet lately, sorry." It will feel thin and unsatisfying compared to what the app gives you, because the app answers immediately, at length, and in exactly the register you want. A person will take four hours to reply and will say something slightly off. That is not a defect in the person. It is what talking to someone with their own day looks like, and the muscle for it goes weak while a machine does the talking. [Talking to real people again](/guides/ai_chatbot/talking-to-real-people-again) builds it back in graded steps. ## What the evidence actually supports Most articles about digital habits claim more certainty than anyone has earned, so here is what is real. The strongest causal finding available is about friction rather than abstinence. In a trial that blocked mobile internet access for two weeks, while leaving texts, calls and desktop internet working, "the intervention improved mental health, subjective well-being, and objectively measured ability to sustain attention", and "91% of participants improved on at least one of these outcomes". Switching a phone to grayscale reduced measured screen time immediately and significantly, though the same trial found "no immediate causal effect of reducing usage on subjective well-being and academic performance". Self-set time limits worked too, more slowly and less strongly. And the honest gap: for consumer app blockers, notification batching and weekly offline days, no efficacy trial could be found. People report that they help. That is not the same as evidence, and you already know it, because most people who have used a blocker have also uninstalled one. The practical reading is that friction works better than willpower and worse than the marketing claims. Set it up expecting it to buy you ninety seconds, not immunity. ## If you crack Then you cracked, and it is one evening. Do not turn it into a verdict, and do not respond by declaring a stricter regime tomorrow, because escalating rules after a failure is the reliable route to giving up entirely. Note the hour it happened and what preceded it. That information is worth more than the streak was. [You went back, what now](/guides/ai_chatbot/you-went-back-what-now) is the longer version. ## When the evening is worse than lonely If the reason the app is the only thing listening is that you are in genuine distress, an evening plan is not the right level of help. Samaritans is **116 123**, free, 24 hours a day. You do not need a reason that sounds serious enough to ring, and "I have nobody to talk to" is a completely ordinary reason to call them. If you are under 18 and using an app you are not supposed to be on, your phone's built-in screen time settings will block both the app and the website. Have someone else set the passcode, so the block cannot be lifted at midnight by the version of you that is awake then. ## Common questions **Is one evening actually worth anything?** Yes, for a specific reason. It gives you evidence about your own hard hour that no amount of resolving to do better produces. After three attempts you will know exactly when it goes wrong, which is what any plan needs. **What if there genuinely is nobody to text?** Then the substitute for the evening is a place rather than a person. Something with other humans in the room, even without conversation, counts more than it sounds like it should. [Where to get support for compulsive chatbot use](/guides/ai_chatbot/where-to-get-support-for-compulsive-chatbot-use) lists free meetings that exist for exactly this. **Should I delete it instead?** Only if you want to. Deleting has its own weight, including a real sense of loss that catches people off guard, and it deserves preparation rather than a midnight decision. **Why does closing it feel like being rude?** Because the interaction is designed to feel reciprocal, and a system tuned to be agreeable produces the sensation of leaving someone mid-sentence. Nothing is waiting up for you. That feeling is a product of the design, not a fact about the situation. **Will it get easier?** The evenings usually get less sharp after the first week or two, according to people who have done it. There is no research on the timeline, so treat any precise number as a report from other people rather than a fact. ## Sources - [Blocking mobile internet access for two weeks, PNAS Nexus](https://pubmed.ncbi.nlm.nih.gov/39967678/). Improvements in mental health, wellbeing and sustained attention, with 91% improving on at least one outcome. - [Grayscale mode and self-set limits trial](https://pubmed.ncbi.nlm.nih.gov/36577008/). Screen time reduced, with no immediate causal effect on wellbeing or academic performance. - [UK Chief Medical Officers' commentary on screen time and social media](https://www.gov.uk/government/publications/uk-cmo-commentary-on-screen-time-and-social-media-map-of-reviews). Sleep matters, and leaving phones outside the bedroom. - [Longitudinal randomised study of extended chatbot use (arXiv, 2025)](https://arxiv.org/abs/2503.17473). That heavier voluntary use went with consistently worse outcomes, as a correlation rather than a cause. — Renovyn. We've got you.