# Courage Horizon > Courage Horizon is an independent technology company. We build tools for the second-order effects of new technology: the slow, hard-to-see changes that shape how people live. Our products help people flourish through those changes. Courage Horizon is based in London. Contact: hello@couragehorizon.com ## Core thesis Every technology arrives twice: first as what it does (first-order effects, visible and widely discussed), then as what it changes (second-order effects: slow shifts in how people relate, focus, learn and trust). People consistently overindex on the first and underindex on the second. Courage Horizon builds paid software tools aimed at those second-order effects, helping people flourish through technological change. ## Tools - [Coppice](https://couragehorizon.com/coppice): Coppice is a private iOS app for looking after your relationships. It holds notes on the people who matter to you and gently reminds you to reach out, so good relationships do not fade through neglect. (iOS app) - [Agni](https://couragehorizon.com/agni): A free three-minute test that scores your understanding of AI from 0 to 100 and shows where you sit against people doing similar work. Agni, the free daily learning app built to raise that score, is on the App Store for iPhone. (iOS app) - [Sourcive](https://couragehorizon.com/sourcive): An AI search visibility audit. See what ChatGPT, Claude, Perplexity and Gemini say when your customers ask for recommendations, and get the files and fix plan to change it. (Web app) ## Guides Practical answers to the questions businesses ask about being seen, understood and recommended by AI assistants. - [How do I get my website recommended by ChatGPT?](https://couragehorizon.com/guides/how-do-i-get-my-website-recommended-by-chatgpt): Publish direct answers to the questions your buyers ask, make your site machine-readable, and get present on the sources AI assistants already cite. - [How do I check what ChatGPT says about my business?](https://couragehorizon.com/guides/how-do-i-check-what-chatgpt-says-about-my-business): Ask the assistants the questions your customers ask, not just your own name, and record whether you appear, how you are described, and who is named instead. - [Why doesn't ChatGPT mention my business?](https://couragehorizon.com/guides/why-doesnt-chatgpt-mention-my-business): Usually one of four fixable reasons: your site never answers buyer questions, machines cannot read it, no trusted source mentions you, or the category answer is already owned. - [Why does ChatGPT describe my company incorrectly?](https://couragehorizon.com/guides/why-does-chatgpt-describe-my-company-incorrectly): The model is filling gaps with whatever it has: stale training data, a similarly named company, or vague guesses from a site that never states the facts plainly. You fix it by publishing the facts where machines can read them. - [What is llms.txt and does my website need one?](https://couragehorizon.com/guides/what-is-llms-txt-and-does-my-website-need-one): A plain-text file at /llms.txt that summarises your site for AI systems. Cheap to add, no downside, and one part of being accurately understood, not a magic ranking switch. - [Why does ChatGPT recommend your competitor instead of you?](https://couragehorizon.com/guides/why-does-chatgpt-recommend-your-competitor-instead-of-you): ChatGPT recommends your competitor because their page is easier for the model to retrieve, trust and quote, not because their product is better. - [Why Won't ChatGPT Say My Business Is Legit?](https://couragehorizon.com/guides/why-won-t-chatgpt-say-my-business-is-legit): ChatGPT hedges on legitimacy questions because a business has not published the specific, checkable evidence a model needs to vouch for it, not because the model doubts the business. - [How Do You Get Your Business Mentioned by AI Tools Like ChatGPT?](https://couragehorizon.com/guides/how-do-you-get-your-business-mentioned-when-people-ask-ai-tools-for-recommendations): A business gets mentioned by AI tools mainly by being cited in third-party discussion, especially Reddit threads, rather than by rewriting its own website. - [Why doesn't ChatGPT recommend my business?](https://couragehorizon.com/guides/why-doesn-t-chatgpt-recommend-my-business): ChatGPT and Perplexity each run separate retrieval and citation systems, so a legitimate business goes unmentioned on one or both because its content doesn't match how buyers phrase the question there, not because anything about the business is wrong. - [Have People Stopped Googling and Started Asking ChatGPT Instead?](https://couragehorizon.com/guides/have-people-actually-stopped-googling-and-started-asking-chatgpt-instead): Yes, for roughly half the country: Pew Research Center found in June 2026 that 49% of US adults now use an AI chatbot and 44% use ChatGPT specifically, up from 33% in 2024. - [What Is AI Share of Voice, and How Do You Actually Measure It?](https://couragehorizon.com/guides/what-is-share-of-voice-in-ai-search-and-how-is-it-actually-measured): AI share of voice is the percentage of AI-generated answers in your category that mention your brand, calculated as (brand mentions / total category mentions across a prompt set) x 100. - [Does robots.txt block ChatGPT? Not the bot that actually cites you](https://couragehorizon.com/guides/does-robots-txt-block-chatgpt-and-other-ai-crawlers): Robots.txt can stop OpenAI's training crawler if you configure it correctly, but it does nothing to stop the retrieval and browsing bots that generate live citations in ChatGPT and Perplexity answers. - [Why Perplexity won't cite your site, even with great backlinks](https://couragehorizon.com/guides/how-do-i-get-my-website-cited-as-a-source-by-perplexity): Perplexity cites pages that are technically crawlable and structured with a direct, quotable answer near the top, not pages with the strongest backlink profile or Google ranking. ## Essays - [The Second Order](https://couragehorizon.com/writing/the-second-order): Every technology arrives twice. First as what it does, then as what it changes. This essay sets out why the largest effects of any new technology are the ones nobody prices in, and why this company exists to build for them. ## Coppice writing Coppice's own writing on staying in touch with the relationships that matter. - [Is it too late to reconnect with an old friend?](https://couragehorizon.com/coppice/writing/is-it-too-late-to-reconnect-with-an-old-friend): No. The discomfort of reaching out is almost always worse in your head than in the reply you get, and one plain, short message is enough to test that. - [How do you know when a friendship is actually over, not just quiet?](https://couragehorizon.com/coppice/writing/how-do-you-know-when-a-friendship-is-actually-over-not-just-quiet): A friendship is quiet when either of you would still reach for it given a reason, and over when neither of you would; check which one is true rather than guessing from the silence alone. - [Is it worth asking ChatGPT for relationship advice?](https://couragehorizon.com/coppice/writing/is-it-worth-asking-chatgpt-for-relationship-advice): For thinking out loud it can help, but for working out what to do about one specific person it can only guess, because it has no independent knowledge of that person. - [Why Do I Keep Meaning to Text People Back and Never Actually Do It?](https://couragehorizon.com/coppice/writing/why-do-i-keep-meaning-to-text-people-back-and-never-do): The gap between meaning to reply and actually replying is a friction problem, not a caring problem, and it closes with a system that resurfaces the intention rather than more guilt. - [Why is it so hard to keep in touch with friends when you have ADHD?](https://couragehorizon.com/coppice/writing/why-is-it-so-hard-to-keep-in-touch-with-friends-when-you-have-adhd): Keeping in touch is hard with ADHD because a friend who isn't currently in view can drop out of working memory entirely, and the fix is an external reminder system, not more effort to remember. - [If we stop talking, how long before we're not really friends anymore?](https://couragehorizon.com/coppice/writing/if-we-stop-talking-how-long-before-we-re-not-really-friends-anymore): A friendship does not fail on a single day of silence; it fades in proportion to how much shared contact has actually stopped, which makes accumulated contact, not months elapsed, the number worth tracking. - [Why do I forget about people I love the moment they're out of sight?](https://couragehorizon.com/coppice/writing/why-do-i-forget-about-people-i-love-the-moment-they-re-out-of-sight): Forgetting to reach out to people you love is usually an attention-cueing problem rather than a caring problem, which means the fix is a better reminder, not a bigger feeling. - [How many close friends does the average adult have?](https://couragehorizon.com/coppice/writing/how-many-friends-does-the-average-adult-actually-have-anymore): Most adults now have between three and five close friends, and the share with none at all has risen sharply over the past three decades, because modern life offers less of the repeated, unplanned contact that turns acquaintances into close friends. - [Can ChatGPT give good relationship advice if it doesn't know the person?](https://couragehorizon.com/coppice/writing/why-doesn-t-chatgpt-s-relationship-advice-actually-help-you-keep-in-touch): No, because ChatGPT generates advice fresh from language patterns each time you ask, with no stored memory of who the specific person is or how the relationship currently stands. - [How often should you contact friends to stay close?](https://couragehorizon.com/coppice/writing/how-often-should-you-contact-friends-to-stay-close): The five people closest to you tend to need contact roughly once a week, your next fifteen need something closer to monthly contact, and going around six months without any contact is roughly the point at which a friendship drifts down a layer, according to evolutionary psychologist Robin Dunbar's research on social group sizes. - [What is ADHD object permanence, and does it really affect friendships?](https://couragehorizon.com/coppice/writing/what-is-adhd-object-permanence-and-does-it-really-affect-friendships): ADHD object permanence is a community-coined term, not a recognised clinical symptom, for forgetting someone exists once they're out of sight, and it can genuinely strain friendships through lapses in working memory. - [Why do I forget to text people back?](https://couragehorizon.com/coppice/writing/why-do-i-forget-to-text-people-back): You forget to text people back mainly because a reply has no fixed shape, so it gets postponed, and the longer it waits the bigger the reply you seem to owe becomes, and for some people, because a message that isn't visible stops registering as unfinished business at all. ## Agni guides Plain answers to the questions people ask about understanding and using AI: how these systems work, where they fail, and what is worth learning first. - [Will AI take my job?](https://couragehorizon.com/agni/learn/is-ai-actually-going-to-take-my-job): AI splits jobs into two paths, professionalised or democratised, and which one is yours depends on your actual tasks rather than your job title. - [Does Relying on ChatGPT Make You Worse at Thinking?](https://couragehorizon.com/agni/learn/is-relying-on-ai-making-me-worse-at-thinking-for-myself): Yes, when it does the composing for you: an MIT Media Lab EEG study tracked 54 people across four months and found ChatGPT-assisted writers had the weakest brain connectivity of three groups, against the strongest, most distributed connectivity in people who wrote unaided. - [Why Do I Get Worse ChatGPT Answers Than My Colleague?](https://couragehorizon.com/agni/learn/why-do-i-get-worse-answers-from-chatgpt-than-my-colleague-does): Because ChatGPT answer quality tracks how a question is phrased, not who asks it, and a missing role, situation or output-format instruction is the specific, checkable gap behind a weaker answer. - [Why does AI sound so confident when it's wrong?](https://couragehorizon.com/agni/learn/why-does-ai-sound-so-confident-when-it-s-wrong): AI sounds confident when it's wrong because most of the benchmarks used to train and grade it score a confident guess and an honest "I don't know" the same way, so guessing gets rewarded and uncertainty does not. - [Is there a test to measure how good you are at using AI?](https://couragehorizon.com/agni/learn/is-there-an-actual-test-for-how-good-you-are-at-using-ai): Yes, a structured AI skill assessment that scores your actual output against a fixed rubric can tell you how good you are at using AI, more reliably than how confident you feel. - [Why does it feel like everyone your age already understands AI?](https://couragehorizon.com/agni/learn/does-everyone-your-age-already-understand-ai-better-than-you-do): LinkedIn shows a curated highlight reel of AI use rather than the real picture, and a June 2026 poll of over 2,000 US and UK professionals found most Gen X and Boomer workers have never used generative AI at all while nearly two-thirds of younger employees are actively teaching them. - [Am I behind on AI, or just anxious about it?](https://couragehorizon.com/agni/learn/how-do-i-know-if-i-m-behind-on-ai-or-just-anxious-about-it): Feeling anxious about AI is a documented emotional response separate from your actual AI skill, and the only way to tell them apart is to measure the skill directly. - [How to Tell If You're Using AI Well, Not Just Often](https://couragehorizon.com/agni/learn/how-to-tell-if-you-re-using-ai-tools-well-not-just-often): You can tell you're using AI well by checking whether your judgement about its output is improving, not by counting how many times a day you open the tool.