{"course":{"slug":"ai-for-small-business","title":"AI for a Small Business","url":"https://addaly.com/learn/ai-for-small-business","level":"beginner","track":"work"},"lesson":{"slug":"one-owner-one-tool","title":"One person owns it, for one month","url":"https://addaly.com/learn/ai-for-small-business/one-owner-one-tool","position":8,"of":88,"minutes":8,"answer":"Competence with these tools is built by one person doing one task repeatedly until they can predict its failures, so adoption in a small business is one person, one tool.","keyIdea":"Competence with these tools is built by one person doing one task repeatedly until they can predict its failures, so adoption in a small business is one person, one tool, one month: with what is learned written into a file that survives that person leaving."},"body":{"format":"markdown","text":"## Skill, not access\n\nIn a business of one to ten people, the constraint is never the licence fee. It is that nobody has used the thing enough to know what it gets wrong.\n\nThat knowledge is not transferable by reading. It comes from doing the same task thirty or forty times until you can predict the failure before it happens: you start to feel where it will over-promise, where it will smooth away the detail that mattered, which of your customers' questions it always mishandles. That prediction is the actual asset, and it is worth more than any tool choice.\n\nWhich is why: **one person, one tool, one task, for one month.** Not four people trying four tools. Four shallow impressions and no judgement.\n\n## Who it should be\n\nNot necessarily the owner, and often not the youngest person on the assumption that they are good with computers. The useful traits are different:\n\n- They do the task now, so they can tell good output from plausible output.\n- They are willing to write down what happened.\n- They are sceptical enough to check and not so sceptical that they abandon it in week one.\n\nThe last one is worth thinking about. Enthusiasts skip the checking because they want it to work. Cynics stop after the first bad answer. The person you want is the one who says \"it did that thing again\" and writes down what the thing was.\n\n## Teach it by sitting next to someone\n\nTwenty minutes, one real task, side by side. Not a course, not a video, not a document.\n\nOpen a real customer message. Show them the facts sheet going in. Show them the first draft. Then (and this is the part that teaches) show them a **bad** output and how you noticed. Show it inventing a delivery time. Show it apologising for something you had not agreed. People learn the shape of the failure much faster than they learn the shape of the success, and the failure is the part that keeps the business safe.\n\nEnd by writing three lines together: what worked, what it got wrong, what to check every time. That is your first procedure, and it took twenty minutes.\n\n## The thing already happening in your business\n\nSomebody on your team is already using AI on work, on their personal account, without telling you. Surveys of workplace use put this well above half in most markets, and the rate among small teams is not lower.\n\nThey are not being dishonest. They are being efficient and slightly unsure whether it is allowed.\n\nBanning it does not stop it; it moves it further out of sight, onto personal accounts, where you have no visibility of what customer data has been pasted and no ability to check the settings. The workable response is the opposite: say what is allowed, provide an account so the use happens somewhere you can see, and ask people to tell you what they are using it for.\n\nThe rule is usually two sentences. Something like: *use it to draft and to sort; never paste customer names, card details, health information or anything from a contract; read everything before it goes out.* That covers most of the real risk and does not require a policy document, though you will write one in module seven.\n\n## Write down what you learn, in one file\n\nA plain document, one page, three sections: prompts that work, mistakes it makes, and rules we follow. Add to it whenever something surprising happens.\n\nThis is boring and it is the difference between a business that has learned something and a business where one person has learned something. If that person leaves, and in a small business people do leave, the file is what stays. Without it, adoption in a small business is one person deep and one resignation from zero, which is one of the most common ways a promising pilot ends.\n\n## When to widen\n\nAfter a month, if the file has content and the task is faster, add the second person or the second task. Not both. And keep the file going, because the second person's mistakes will not be the same as the first person's, and that is information about the tool rather than about them."},"check":{"question":"An owner discovers three staff are already using personal AI accounts for work, including pasting customer emails. What is the reasoning behind providing accounts and a two-sentence rule rather than banning the practice?","options":[{"text":"Staff are entitled to choose their own tools, so a ban would damage morale more than it protects data.","correct":false,"tempting":"Argues from staff autonomy rather than from risk, and would justify the same conclusion even where the data being pasted was genuinely dangerous."},{"text":"Personal accounts are usually free tiers, which are cheaper for the business than paid seats.","correct":false,"tempting":"Optimises for cost when the issue is exposure; free consumer tiers are precisely the ones most likely to carry weaker data terms."},{"text":"Provided accounts have configurable settings the owner can check, whereas a ban moves the same pasting onto accounts nobody can see or audit.","correct":true},{"text":"A ban is unenforceable in a small team because there is no IT department to monitor compliance, and a rule nobody can check teaches staff that the business's policies are decorative.","correct":false,"tempting":"Treats enforceability as the whole argument; even a perfectly enforced ban would be worse, because it removes the visibility that makes the risk manageable."}]},"prev":{"slug":"jobs-it-cannot-do","title":"Seven things it will fail at, and why","url":"https://addaly.com/learn/ai-for-small-business/jobs-it-cannot-do"},"next":{"slug":"why-pilots-die","title":"Five ways a small-business pilot dies","url":"https://addaly.com/learn/ai-for-small-business/why-pilots-die"},"authorship":"Founder and series editor: Dr Mustafa Mun. Written by AI models to a written standard. Exam keys checked blind by a second model: 3,665 of 3,666 agreed. Academic review invited.","review":{"status":"Draft, open for academic review"},"cite":"Addaly (2026) 'One person owns it, for one month', in AI for a Small Business. Edited by M. Mun. Addaly. Available at: https://addaly.com/learn/ai-for-small-business/one-owner-one-tool (Accessed: 13 October 2026).","licence":{"name":"CC BY-NC-SA 4.0","url":"https://creativecommons.org/licenses/by-nc-sa/4.0/"},"reader":{"schema":1,"title":"One person owns it, for one month","courseTitle":"AI for a Small Business","module":{"index":1,"slug":"where-it-belongs","title":"Working out where AI belongs in your business"},"answer":null,"keyIdea":"Competence with these tools is built by one person doing one task repeatedly until they can predict its failures, so adoption in a small business is one person, one tool, one month: with what is learned written into a file that survives that person leaving.","hero":{"kind":"plate","image":null,"alt":null,"caption":null},"blocks":[{"t":"h2","id":"skill-not-access","runs":[{"text":"Skill, not access"}]},{"t":"p","runs":[{"text":"In a business of one to ten people, the constraint is never the licence fee. It is that nobody has used the thing enough to know what it gets wrong."}]},{"t":"p","runs":[{"text":"That knowledge is not transferable by reading. It comes from doing the same task thirty or forty times until you can predict the failure before it happens: you start to feel where it will over-promise, where it will smooth away the detail that mattered, which of your customers’ questions it always mishandles. That prediction is the actual asset, and it is worth more than any tool choice."}]},{"t":"p","runs":[{"text":"Which is why: "},{"text":"one person, one tool, one task, for one month.","b":true},{"text":" Not four people trying four tools. Four shallow impressions and no judgement."}]},{"t":"h2","id":"who-it-should-be","runs":[{"text":"Who it should be"}]},{"t":"p","runs":[{"text":"Not necessarily the owner, and often not the youngest person on the assumption that they are good with computers. The useful traits are different:"}]},{"t":"ul","items":[[{"text":"They do the task now, so they can tell good output from plausible output."}],[{"text":"They are willing to write down what happened."}],[{"text":"They are sceptical enough to check and not so sceptical that they abandon it in week one."}]]},{"t":"p","runs":[{"text":"The last one is worth thinking about. Enthusiasts skip the checking because they want it to work. Cynics stop after the first bad answer. The person you want is the one who says “it did that thing again” and writes down what the thing was."}]},{"t":"h2","id":"teach-it-by-sitting-next-to-someone","runs":[{"text":"Teach it by sitting next to someone"}]},{"t":"p","runs":[{"text":"Twenty minutes, one real task, side by side. Not a course, not a video, not a document."}]},{"t":"p","runs":[{"text":"Open a real customer message. Show them the facts sheet going in. Show them the first draft. Then (and this is the part that teaches) show them a "},{"text":"bad","b":true},{"text":" output and how you noticed. Show it inventing a delivery time. Show it apologising for something you had not agreed. People learn the shape of the failure much faster than they learn the shape of the success, and the failure is the part that keeps the business safe."}]},{"t":"p","runs":[{"text":"End by writing three lines together: what worked, what it got wrong, what to check every time. That is your first procedure, and it took twenty minutes."}]},{"t":"h2","id":"the-thing-already-happening-in-your-business","runs":[{"text":"The thing already happening in your business"}]},{"t":"p","runs":[{"text":"Somebody on your team is already using AI on work, on their personal account, without telling you. Surveys of workplace use put this well above half in most markets, and the rate among small teams is not lower."}]},{"t":"p","runs":[{"text":"They are not being dishonest. They are being efficient and slightly unsure whether it is allowed."}]},{"t":"p","runs":[{"text":"Banning it does not stop it; it moves it further out of sight, onto personal accounts, where you have no visibility of what customer data has been pasted and no ability to check the settings. The workable response is the opposite: say what is allowed, provide an account so the use happens somewhere you can see, and ask people to tell you what they are using it for."}]},{"t":"p","runs":[{"text":"The rule is usually two sentences. Something like: "},{"text":"use it to draft and to sort; never paste customer names, card details, health information or anything from a contract; read everything before it goes out.","i":true},{"text":" That covers most of the real risk and does not require a policy document, though you will write one in module seven."}]},{"t":"h2","id":"write-down-what-you-learn-in-one-file","runs":[{"text":"Write down what you learn, in one file"}]},{"t":"p","runs":[{"text":"A plain document, one page, three sections: prompts that work, mistakes it makes, and rules we follow. Add to it whenever something surprising happens."}]},{"t":"p","runs":[{"text":"This is boring and it is the difference between a business that has learned something and a business where one person has learned something. If that person leaves, and in a small business people do leave, the file is what stays. Without it, adoption in a small business is one person deep and one resignation from zero, which is one of the most common ways a promising pilot ends."}]},{"t":"h2","id":"when-to-widen","runs":[{"text":"When to widen"}]},{"t":"p","runs":[{"text":"After a month, if the file has content and the task is faster, add the second person or the second task. Not both. And keep the file going, because the second person’s mistakes will not be the same as the first person’s, and that is information about the tool rather than about them."}]}],"check":{"question":[{"text":"An owner discovers three staff are already using personal AI accounts for work, including pasting customer emails. What is the reasoning behind providing accounts and a two-sentence rule rather than banning the practice?"}],"options":[{"text":[{"text":"Staff are entitled to choose their own tools, so a ban would damage morale more than it protects data."}],"correct":false,"tempting":[{"text":"Argues from staff autonomy rather than from risk, and would justify the same conclusion even where the data being pasted was genuinely dangerous."}]},{"text":[{"text":"Personal accounts are usually free tiers, which are cheaper for the business than paid seats."}],"correct":false,"tempting":[{"text":"Optimises for cost when the issue is exposure; free consumer tiers are precisely the ones most likely to carry weaker data terms."}]},{"text":[{"text":"Provided accounts have configurable settings the owner can check, whereas a ban moves the same pasting onto accounts nobody can see or audit."}],"correct":true,"tempting":null},{"text":[{"text":"A ban is unenforceable in a small team because there is no IT department to monitor compliance, and a rule nobody can check teaches staff that the business’s policies are decorative."}],"correct":false,"tempting":[{"text":"Treats enforceability as the whole argument; even a perfectly enforced ban would be worse, because it removes the visibility that makes the risk manageable."}]}],"order":[3,0,1,2],"why":null},"prev":{"slug":"jobs-it-cannot-do","title":"Seven things it will fail at, and why","minutes":9},"next":{"slug":"why-pilots-die","title":"Five ways a small-business pilot dies","minutes":8}}}