{"course":{"slug":"prompting","title":"Getting Real Work out of a Model","url":"https://addaly.com/learn/prompting","level":"beginner","track":"start"},"lesson":{"slug":"plan-then-execute","title":"Ask for the plan before the work","url":"https://addaly.com/learn/prompting/plan-then-execute","position":42,"of":72,"minutes":8,"answer":"Get the outline and the assumptions that would change the result before any substantial output is produced, and for anything that acts on the world, require the plan to state what is irreversible.","keyIdea":"Get the outline and the assumptions that would change the result before any substantial output is produced, and for anything that acts on the world, require the plan to state what is irreversible."},"body":{"format":"markdown","text":"## The cheapest correction is the one before the work\n\nYou ask for a 2,000-word article, a migration script, a full analysis. Four minutes later you have something substantial built on an assumption you would have rejected in five seconds.\n\nAsk for the plan first.\n\n> Before writing anything: give me the outline as six headings with one line each on what goes under them, and list the three assumptions you are making about the audience and scope. Do not write the article yet.\n\nYou read fifteen lines. You correct the two that are wrong. Then you say go.\n\nThe cost of correcting an assumption rises steeply with how much has been built on it, and this is the move that catches it at the bottom of that curve.\n\n> Figure: What a wrong assumption costs, by when it is caught. Line chart of Minutes lost against Stage at which the assumption is caught: A 2,000-word piece built on the wrong reader (0, 0.5) (1, 2) (2, 8) (3, 15) (4, 60). Stage 0 is the assumptions list, 1 the outline, 2 half a draft, 3 the full draft, 4 after it was sent. The curve is why fifteen lines of plan are worth reading: they catch the error at the bottom of it.\n\n## Ask for the assumptions explicitly\n\nThe outline is useful. The **assumptions list** is more useful, and almost nobody asks for it.\n\nEvery substantial request leaves things unstated, and the model fills them silently: who the reader is, what the timeframe is, whether the figures are annual or monthly, whether \"customers\" means accounts or people. A plan surfaces those as a list you can scan.\n\n> List the assumptions you are making that, if wrong, would change the result.\n\nThat phrasing is deliberate. It filters out trivia and produces the three or four that actually matter.\n\n## Where plan-first pays most\n\n**Long outputs.** Anything over about 500 words. The cost of a wrong outline is the whole draft.\n\n**Anything that will be run.** A script, a spreadsheet formula, a bulk change. Read the plan before something touches real data.\n\n**Anything with a structure you will argue about.** Reports, proposals, curricula. Structure is where most of the disagreement lives, and it is much cheaper to argue about six headings than about 2,000 words.\n\n**Multi-step work with tools.** If a model is going to search, read files or call things, the plan tells you what it is about to do while it is still a sentence.\n\n## Where it is a waste\n\nShort tasks. A 150-word email does not need an outline; asking for one doubles the interaction for no gain. Anything you will read in full and can judge instantly is faster to just produce.\n\n## Approving a plan is a real decision\n\nTwo failure modes to watch in yourself.\n\n**Rubber-stamping.** A plan that looks reasonable usually is reasonable, so it becomes habit to say \"yes, go\". Read the assumptions list, at least. That is where the surprise is.\n\n**Over-planning.** Some people iterate on the outline four times and produce a beautiful plan for a document nobody needed. Two rounds is almost always enough; if the third round is still finding structural problems, the request itself is unclear and no plan will fix it.\n\n## The version for automated work\n\nWhen a model has tools (running code, editing files, sending things) plan-first stops being a convenience and becomes a control.\n\n> Produce the plan as a numbered list of actions. For each, say what it changes and whether it is reversible. Wait for my approval. Do not act.\n\nThe reversibility column is the useful one. It sorts the actions into those you can skim and those you must read, and it is the shape of every sensible approval gate in automated systems. Anything irreversible (sending, deleting, paying, publishing) should have a human between the plan and the act, and that principle does not weaken as models improve.\n\n## The two-line habit\n\nFor anything substantial:\n\n> First: outline plus the assumptions that would change the result if wrong. Then stop.\n\nFifteen seconds to type, and it will save you an afternoon roughly once a fortnight.\n\n## A worked case\n\nA consultant asked for a 3,000-word market overview and got something competent about the wrong market: the model had read \"the education sector\" as schools when she meant corporate training. She noticed on page two.\n\nThe plan version would have surfaced it in one line: *Assumption: \"education sector\" means K-12 and higher education institutions.* Ten seconds to read, five seconds to correct, and the twelve minutes of generation would have been spent on the right subject.\n\nThe interesting part is that her prompt was not badly written. It was ambiguous in a way that is invisible from the inside, because she knew what she meant. That is the normal case, not a lapse, and it is why the assumptions list beats a careful re-read of your own prompt."},"check":{"question":"Which request most reliably surfaces the problem before a long draft is written?","options":[{"text":"\"Give me a detailed outline with subheadings and estimated word counts.\"","correct":false,"tempting":"Detailed structure without surfacing the unstated premises the structure rests on."},{"text":"\"Confirm you have understood my request before proceeding.\"","correct":false,"tempting":"Invites a restatement, which a model will produce agreeably whether or not it has understood."},{"text":"\"Give me the outline, plus the assumptions you are making that would change the result if they were wrong.\"","correct":true},{"text":"\"Ask me any clarifying questions you have before writing.\"","correct":false,"tempting":"Genuinely useful and covered in a later lesson, but questions surface what the model noticed as uncertain, not the assumptions it has silently settled."}]},"prev":{"slug":"breaking-a-task-into-steps","title":"Splitting a task that keeps failing","url":"https://addaly.com/learn/prompting/breaking-a-task-into-steps"},"next":{"slug":"checking-against-constraints","title":"Making it check its answer against the rules","url":"https://addaly.com/learn/prompting/checking-against-constraints"},"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) 'Ask for the plan before the work', in Getting Real Work out of a Model. Edited by M. Mun. Addaly. Available at: https://addaly.com/learn/prompting/plan-then-execute (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":"Ask for the plan before the work","courseTitle":"Getting Real Work out of a Model","module":{"index":5,"slug":"making-it-work-the-problem","title":"Making it work the problem"},"answer":null,"keyIdea":"Get the outline and the assumptions that would change the result before any substantial output is produced, and for anything that acts on the world, require the plan to state what is irreversible.","hero":{"kind":"figure","figure":{"t":"figure","label":"Figure","figure":{"type":"line","title":"What a wrong assumption costs, by when it is caught","series":[{"label":"A 2,000-word piece built on the wrong reader","points":[[0,0.5],[1,2],[2,8],[3,15],[4,60]]}],"xLabel":"Stage at which the assumption is caught","yLabel":"Minutes lost","caption":"Stage 0 is the assumptions list, 1 the outline, 2 half a draft, 3 the full draft, 4 after it was sent. The curve is why fifteen lines of plan are worth reading: they catch the error at the bottom of it."},"alt":"Figure: What a wrong assumption costs, by when it is caught. Line chart of Minutes lost against Stage at which the assumption is caught: A 2,000-word piece built on the wrong reader (0, 0.5) (1, 2) (2, 8) (3, 15) (4, 60). Stage 0 is the assumptions list, 1 the outline, 2 half a draft, 3 the full draft, 4 after it was sent. The curve is why fifteen lines of plan are worth reading: they catch the error at the bottom of it."}},"blocks":[{"t":"h2","id":"the-cheapest-correction-is-the-one-before-the-work","runs":[{"text":"The cheapest correction is the one before the work"}]},{"t":"p","runs":[{"text":"You ask for a 2,000-word article, a migration script, a full analysis. Four minutes later you have something substantial built on an assumption you would have rejected in five seconds."}]},{"t":"p","runs":[{"text":"Ask for the plan first."}]},{"t":"quote","runs":[{"text":"Before writing anything: give me the outline as six headings with one line each on what goes under them, and list the three assumptions you are making about the audience and scope. Do not write the article yet."}]},{"t":"p","runs":[{"text":"You read fifteen lines. You correct the two that are wrong. Then you say go."}]},{"t":"p","runs":[{"text":"The cost of correcting an assumption rises steeply with how much has been built on it, and this is the move that catches it at the bottom of that curve."}]},{"t":"h2","id":"ask-for-the-assumptions-explicitly","runs":[{"text":"Ask for the assumptions explicitly"}]},{"t":"p","runs":[{"text":"The outline is useful. The "},{"text":"assumptions list","b":true},{"text":" is more useful, and almost nobody asks for it."}]},{"t":"p","runs":[{"text":"Every substantial request leaves things unstated, and the model fills them silently: who the reader is, what the timeframe is, whether the figures are annual or monthly, whether “customers” means accounts or people. A plan surfaces those as a list you can scan."}]},{"t":"quote","runs":[{"text":"List the assumptions you are making that, if wrong, would change the result."}]},{"t":"p","runs":[{"text":"That phrasing is deliberate. It filters out trivia and produces the three or four that actually matter."}]},{"t":"h2","id":"where-plan-first-pays-most","runs":[{"text":"Where plan-first pays most"}]},{"t":"p","runs":[{"text":"Long outputs.","b":true},{"text":" Anything over about 500 words. The cost of a wrong outline is the whole draft."}]},{"t":"p","runs":[{"text":"Anything that will be run.","b":true},{"text":" A script, a spreadsheet formula, a bulk change. Read the plan before something touches real data."}]},{"t":"p","runs":[{"text":"Anything with a structure you will argue about.","b":true},{"text":" Reports, proposals, curricula. Structure is where most of the disagreement lives, and it is much cheaper to argue about six headings than about 2,000 words."}]},{"t":"p","runs":[{"text":"Multi-step work with tools.","b":true},{"text":" If a model is going to search, read files or call things, the plan tells you what it is about to do while it is still a sentence."}]},{"t":"h2","id":"where-it-is-a-waste","runs":[{"text":"Where it is a waste"}]},{"t":"p","runs":[{"text":"Short tasks. A 150-word email does not need an outline; asking for one doubles the interaction for no gain. Anything you will read in full and can judge instantly is faster to just produce."}]},{"t":"h2","id":"approving-a-plan-is-a-real-decision","runs":[{"text":"Approving a plan is a real decision"}]},{"t":"p","runs":[{"text":"Two failure modes to watch in yourself."}]},{"t":"p","runs":[{"text":"Rubber-stamping.","b":true},{"text":" A plan that looks reasonable usually is reasonable, so it becomes habit to say “yes, go”. Read the assumptions list, at least. That is where the surprise is."}]},{"t":"p","runs":[{"text":"Over-planning.","b":true},{"text":" Some people iterate on the outline four times and produce a beautiful plan for a document nobody needed. Two rounds is almost always enough; if the third round is still finding structural problems, the request itself is unclear and no plan will fix it."}]},{"t":"h2","id":"the-version-for-automated-work","runs":[{"text":"The version for automated work"}]},{"t":"p","runs":[{"text":"When a model has tools (running code, editing files, sending things) plan-first stops being a convenience and becomes a control."}]},{"t":"quote","runs":[{"text":"Produce the plan as a numbered list of actions. For each, say what it changes and whether it is reversible. Wait for my approval. Do not act."}]},{"t":"p","runs":[{"text":"The reversibility column is the useful one. It sorts the actions into those you can skim and those you must read, and it is the shape of every sensible approval gate in automated systems. Anything irreversible (sending, deleting, paying, publishing) should have a human between the plan and the act, and that principle does not weaken as models improve."}]},{"t":"h2","id":"the-two-line-habit","runs":[{"text":"The two-line habit"}]},{"t":"p","runs":[{"text":"For anything substantial:"}]},{"t":"quote","runs":[{"text":"First: outline plus the assumptions that would change the result if wrong. Then stop."}]},{"t":"p","runs":[{"text":"Fifteen seconds to type, and it will save you an afternoon roughly once a fortnight."}]},{"t":"h2","id":"a-worked-case","runs":[{"text":"A worked case"}]},{"t":"p","runs":[{"text":"A consultant asked for a 3,000-word market overview and got something competent about the wrong market: the model had read “the education sector” as schools when she meant corporate training. She noticed on page two."}]},{"t":"p","runs":[{"text":"The plan version would have surfaced it in one line: "},{"text":"Assumption: “education sector” means K-12 and higher education institutions.","i":true},{"text":" Ten seconds to read, five seconds to correct, and the twelve minutes of generation would have been spent on the right subject."}]},{"t":"p","runs":[{"text":"The interesting part is that her prompt was not badly written. It was ambiguous in a way that is invisible from the inside, because she knew what she meant. That is the normal case, not a lapse, and it is why the assumptions list beats a careful re-read of your own prompt."}]}],"check":{"question":[{"text":"Which request most reliably surfaces the problem before a long draft is written?"}],"options":[{"text":[{"text":"“Give me a detailed outline with subheadings and estimated word counts.”"}],"correct":false,"tempting":[{"text":"Detailed structure without surfacing the unstated premises the structure rests on."}]},{"text":[{"text":"“Confirm you have understood my request before proceeding.”"}],"correct":false,"tempting":[{"text":"Invites a restatement, which a model will produce agreeably whether or not it has understood."}]},{"text":[{"text":"“Give me the outline, plus the assumptions you are making that would change the result if they were wrong.”"}],"correct":true,"tempting":null},{"text":[{"text":"“Ask me any clarifying questions you have before writing.”"}],"correct":false,"tempting":[{"text":"Genuinely useful and covered in a later lesson, but questions surface what the model noticed as uncertain, not the assumptions it has silently settled."}]}],"order":[3,0,2,1],"why":null},"prev":{"slug":"breaking-a-task-into-steps","title":"Splitting a task that keeps failing","minutes":9},"next":{"slug":"checking-against-constraints","title":"Making it check its answer against the rules","minutes":8}}}