--- title: "Write a social media context file so AI posts sound like your account" description: "Store voice rules, content pillars, platform rules, post structures, and your best posts in one file, so AI social drafts sound like you." canonical: "https://scalewithsearch.com/articles/ai-memory-for-social-media" date: "2026-01-28" modified: "2026-09-25" --- ## Site navigation - [Scale With Search](https://scalewithsearch.com/) - Real estate - Real estate - [Real estate](https://scalewithsearch.com/for/real-estate) - Work - Start here - [Send your brief](https://scalewithsearch.com/work#send-your-brief) - [Prepare your six-question brief](https://scalewithsearch.com/work#prepare-your-six-question-brief) - Build - [Site build, content library with SEO, signal desk](https://scalewithsearch.com/work) - For your business - Trades and home services - [Auto body and collision shops](https://scalewithsearch.com/for/auto-body-and-collision-shops) - [Foundation and home repair contractors](https://scalewithsearch.com/for/foundation-and-home-repair) - [Garage door and fencing contractors](https://scalewithsearch.com/for/garage-door-and-fencing-contractors) - [HVAC contractors](https://scalewithsearch.com/for/hvac-contractors) - [Janitorial and commercial cleaning companies](https://scalewithsearch.com/for/janitorial-and-commercial-cleaning) - [Locksmiths](https://scalewithsearch.com/for/locksmiths) - [Moving companies](https://scalewithsearch.com/for/moving-companies) - [Pest control companies](https://scalewithsearch.com/for/pest-control-companies) - [Plumbing and electrical contractors](https://scalewithsearch.com/for/plumbing-and-electrical-contractors) - [Restoration and water or fire damage companies](https://scalewithsearch.com/for/restoration-and-water-fire-damage) - [Roofing companies](https://scalewithsearch.com/for/roofing-companies) - [Towing companies](https://scalewithsearch.com/for/towing-companies) - [Tree services and landscaping companies](https://scalewithsearch.com/for/tree-services-and-landscaping) - [Solar installers](https://scalewithsearch.com/for/solar-installation) - [General contractors](https://scalewithsearch.com/for/general-contractors-and-construction) - [Paving, concrete, and flooring contractors](https://scalewithsearch.com/for/paving) - Practices and professional services - [Bookkeeping and tax practices](https://scalewithsearch.com/for/bookkeeping-and-tax-practices) - [Dental practices](https://scalewithsearch.com/for/dental-practices) - [Family and criminal defense law firms](https://scalewithsearch.com/for/family-and-criminal-defense-law-firms) - [Med spas and aesthetics practices](https://scalewithsearch.com/for/med-spas-and-aesthetics) - [Personal injury law firms](https://scalewithsearch.com/for/personal-injury-law-firms) - [Veterinary clinics](https://scalewithsearch.com/for/veterinary-clinics) - [Gyms and fitness studios](https://scalewithsearch.com/for/fitness) - [Therapy and outpatient health practices](https://scalewithsearch.com/for/therapy-and-outpatient-health) - [Medical billing companies](https://scalewithsearch.com/for/medical-billing) - [Insurance agencies](https://scalewithsearch.com/for/insurance-agencies) - [Financial advisors](https://scalewithsearch.com/for/financial-advisors) - [Property management companies](https://scalewithsearch.com/for/property-management) - [Recruiting and staffing agencies](https://scalewithsearch.com/for/recruiting-and-staffing) - [Architects and interior designers](https://scalewithsearch.com/for/architects-and-interior-designers) - [Logistics and supply chain companies](https://scalewithsearch.com/for/logistics-and-supply-chain) - Agencies, MSPs, and manufacturing - [IT and managed service providers](https://scalewithsearch.com/for/it-and-managed-service-providers) - [Machine shops and precision manufacturers](https://scalewithsearch.com/for/machine-shops-and-precision-manufacturing) - [Marketing agencies and freelancers](https://scalewithsearch.com/for/marketing-agencies-and-freelancers) - [SEO agencies and consultants](https://scalewithsearch.com/for/seo-agencies-and-consultants) - [Small manufacturers and fabricators](https://scalewithsearch.com/for/small-manufacturers-and-fabricators) - Restaurants, shops, studios, and nonprofits - [Restaurants and hospitality businesses](https://scalewithsearch.com/for/restaurants-and-hospitality) - [Retail stores and ecommerce sellers](https://scalewithsearch.com/for/retail-and-ecommerce) - [Photographers, event planners, and travel agents](https://scalewithsearch.com/for/photographers) - [Churches and nonprofits](https://scalewithsearch.com/for/churches-and-nonprofits) - [All industries](https://scalewithsearch.com/for/) - Learn - For your office - [Office job guides](https://scalewithsearch.com/guides/) - [Browser calculators](https://scalewithsearch.com/tools/) - Start here - [How it works](https://scalewithsearch.com/how-it-works) - [Free Starter Kit](https://scalewithsearch.com/kit/business-memory-starter-kit.zip) - [Synthetic specimen](https://scalewithsearch.com/specimen/working-session-specimen.zip) - Guides - [The Complete Guide to Business Memory for AI Agents](https://scalewithsearch.com/articles/business-memory-for-ai-agents-guide) - [The Complete Small-Business Guide to AI Agent Governance](https://scalewithsearch.com/articles/ai-agent-governance-guide-small-business) - [The Complete Guide to Leaving Vendor AI Memory](https://scalewithsearch.com/articles/leaving-vendor-ai-memory-guide) - Articles by cluster - [Business memory](https://scalewithsearch.com/articles/business-memory-for-ai-agents-guide) - [Agent governance](https://scalewithsearch.com/articles/ai-agent-governance-guide-small-business) - [Migration and ownership](https://scalewithsearch.com/articles/leaving-vendor-ai-memory-guide) - For machines - [llms.txt](https://scalewithsearch.com/llms.txt) - [llms-full.txt](https://scalewithsearch.com/llms-full.txt) - [Machine view](https://scalewithsearch.com/?view=machine) - Company - Evidence - [Proof](https://scalewithsearch.com/proof) - Company - [About](https://scalewithsearch.com/about) # Write a social media context file so AI posts sound like your account. You ask an AI assistant for LinkedIn posts. It returns polished posts that sound like every other post in your feed: "🚀 Excited to share this insight about [Topic]. Here are 3 key takeaways: [Obvious Point 1], [Obvious Point 2], [Obvious Point 3]. What are your thoughts? Drop a comment below! 💡" The post has no personality and no voice. A reader cannot connect it to you. The assistant defaults to generic output because it does not know your account. It does not know your content pillars, your voice, your audience, your best posts, or your platform rules. A social media context file gives it those. It is one Markdown file, `content.md`, that the assistant reads before it drafts. The performance figures below are synthetic examples. ## Know what the assistant needs before it drafts Take the request: "Write a LinkedIn post about the proposal automation system we built. Technical audience, efficiency pillar." A post in your voice needs six things: - your content pillars and when to use each one; - your voice rules: contractions, sentence length, banned phrases; - your formatting rules for each platform; - the post structures that got engagement; - your audience segments and what each one cares about; - examples of posts that performed. With these in the file, the draft follows your structure and fits the platform. You check it and decide whether it goes out. ## Write voice rules the assistant can follow "Be conversational" is not a rule. Write rules that a reader could check on the finished post: ```markdown ## Voice rules DO: - Use contractions (you're, don't, it's). - Start a sentence with "And" or "But" when it reads naturally. - Write short sentences. Mix in one-sentence paragraphs. - Ask at most one question per post. - Use second person (you, your). - Give one concrete example for each abstract point. - Use parentheses for asides (like this). DON'T: - Emoji spam (one per post at most, only if natural). - Corporate jargon (synergy, leverage, solutions, ecosystem). - Hashtag stuffing (3 at most on LinkedIn, none on X). - Exclamation points for emphasis. Use a specific fact instead. - A question in the first line unless you really ask the audience. - "Excited to share" or "Thrilled to announce." - Lists of obvious takeaways. Sentence length: 10 to 20 words on average; mix in 5-word and 30-word sentences. Paragraph length: 1 to 3 sentences. Post length: 100 to 150 words on LinkedIn; 200 to 280 characters on X. ``` Each rule is a check. A search for "Excited to share" either finds the phrase or does not. For how voice rules sit beside claims that change over time, see [AI brand memory for content creation](/articles/ai-brand-memory-content-creation). ## Define content pillars with an angle and a proof format A pillar is a topic plus the angle you take on it, the audience it serves, and the kind of proof it uses. ```markdown ## Content pillars ### 1. Efficiency systems (40% of posts) what:: automation, workflows, time-saving processes angle:: specific systems with before and after time figures audience:: operators and founders buried in busy work topics:: proposal automation, meeting note systems, email workflows proof:: "Used to take 2 hours, now takes 20 minutes" ### 2. AI implementation (30% of posts) what:: how to use AI tools in practice, not theory angle:: practical walkthroughs; what works and what does not audience:: people who tried AI and are frustrated with it topics:: context file setup, prompt design, tool comparisons proof:: screenshots, exact prompts, real examples ### 3. Business outcomes (20% of posts) what:: revenue effect, client results, measurable wins angle:: case studies with real numbers audience:: decision makers and budget holders topics:: client results, return breakdowns, process improvements proof:: cost saved in a stated time, or a percentage change with its source ### 4. Industry commentary (10% of posts) what:: reactions to AI news, trends, predictions angle:: the point others miss audience:: early adopters tracking the AI space topics:: tool releases, hype cycles, practical effects proof:: your own experience, not speculation ``` The percentages tell the assistant how to balance a week of posts. Without them, it writes whichever pillar the request mentions and drifts toward commentary. ## Describe each audience segment by the questions it asks A segment is a group of readers who ask the same questions. Name each segment by role, not by age or location. Then copy the questions that segment asks in comments and direct messages, in its own words. The questions tell the assistant what a post must answer. ```markdown ## Audience segments ### Operators role:: owners and operations leads at firms of 5 to 50 people asks:: "How long does it take to set up?" "What breaks first?" wants:: steps, time figures, the failure case avoid:: tool comparisons that end without a recommendation ### Technical buyers role:: developers and IT leads who evaluate tools asks:: "What does it run on?" "Where does the data go?" wants:: exact commands, file names, limits avoid:: outcome claims without the method ### Decision makers role:: founders and budget holders asks:: "How many hours does it save?" "What is the risk?" wants:: one before and after figure with its source avoid:: jargon, and tool names without a reason ``` Go back to the request in the first section. It asked for a "technical audience". With this block, the assistant knows that this reader wants the file names and the data flow. Without the block, it guesses, and the guess is usually a list of benefits. Keep the questions current. Once a month, read the comments on your last ten posts. Add any question that came up twice, and delete any question that nobody asked in three months. ## Record platform rules from your own results LinkedIn, X, and Instagram differ in length, structure, and what readers expect. Platform ranking also changes without notice. Write down the rules your own account follows, with the date you last checked each one. The block below is one account's rule set from January 2026. ```markdown ## Platform rules (last checked 2026-01) ### LinkedIn - length: 100 to 150 words; 200 at most for a deep dive - structure: hook line, line break, 2 or 3 short paragraphs, optional call to action - hook: the first sentence must stand alone; it shows before "see more" - format: a line break between paragraphs; no wall of text - links: in the first comment, not in the post body (our reach dropped with links in the body) - hashtags: 2 or 3, relevant only - times: Tuesday to Thursday, 8 to 10am or 12 to 1pm ET - engagement: reply to every comment within 2 hours ### X - length: 200 to 280 characters; a thread for longer ideas - structure: one direct statement, or a setup and a payoff - threads: each post stands alone; 4 to 7 posts - format: no line break in the middle of a sentence - links: include when relevant (no reach drop observed on our account) - hashtags: none, unless the topic is trending - times: weekday mornings 7 to 9am ET; lunch 12 to 1pm ET - engagement: quote or reply to related posts; do not only broadcast ### Instagram (if used) - a visual is required; no text-only posts - caption: 50 to 100 words, hook in the first line - hashtags: 5 to 10 relevant tags - stories: behind the scenes, less polished ``` The parenthetical notes matter. "Our reach dropped with links in the body" is an observation you can recheck. "The algorithm hates links" is a rumor. ## Store the post structures that worked After a few hundred posts, you know which structures earn engagement. Store each one with its pattern and its measured result. ```markdown ## Post structures ### Before and after (efficiency pillar) - line 1: "Used to spend X hours on [task]." - line 2: "Now it takes X minutes." - line break - what changed: the system, tool, or process - one specific example - optional: a metric or the next step performance:: 3.2% average engagement; many saves ### Specific problem, specific solution (AI implementation pillar) - the problem in one sentence - why it happens: the cause, not the symptom - the solution, in concrete terms - how to do it, in 2 or 3 steps - the result performance:: 2.8% engagement; high comment rate ### Contrarian point (industry commentary pillar) - the popular opinion, stated fairly - "But here's what they're missing:" - your counterpoint, with proof - why it matters performance:: 4.1% engagement; polarizing (many shares or none) ``` ## Keep a library of top posts with the reason each worked Save your best posts in full, with the numbers and a one-line lesson: ```markdown ## Top posts ### "AI Doesn't Have Amnesia" (2026-01-15, LinkedIn) engagement:: 4.2% rate, 37 comments, 89 shares pillar:: AI implementation why:: a specific pain (AI forgets context), a clear fix (context files), a concrete example full_text:: [paste the full post] lesson:: readers respond to "here's the actual problem" plus "here's the actual fix" ### "Meeting Note System" (2026-01-08, LinkedIn) engagement:: 3.8% rate, 28 comments, 56 shares pillar:: efficiency systems why:: a before and after time figure, a specific tool stack, a familiar pain full_text:: [paste the full post] lesson:: time saved plus specific numbers gets saves ``` The lesson line is what the assistant uses. The full text shows it what the lesson looks like in your words. Label each saved post as structure and voice only. The assistant then does not reuse an old claim as a current fact. ## Compare a draft with and without the file Without the file, "Write a LinkedIn post about using AI for proposal writing" returns this: ```text 🚀 Exciting news in the world of AI and productivity! Are you still spending hours on proposal writing? AI tools can transform your workflow and help you work smarter, not harder. Here are 3 key benefits of using AI for proposals: ✅ Save time on repetitive tasks ✅ Maintain consistency across documents ✅ Focus on high-value client relationships The future of work is here, and it's powered by AI. What are your thoughts on AI in business? Drop a comment below! 💡 #AI #Productivity #BusinessGrowth #Innovation ``` It breaks five rules in the file: emoji spam, the takeaway list, exclamation points, two questions, and four hashtags. With the file, "Write a LinkedIn post about proposal automation; efficiency pillar; before and after format" returns this: ```text Used to spend 2 hours writing proposals. Now it takes 20 minutes. Built a context file with my service packages, case studies, and proposal structure. AI generates first drafts that match my format. I review for accuracy, adjust timeline details, send. Most of my time is checking details, not writing from scratch. Every proposal follows the same structure. Case studies get picked based on client industry. Pricing stays consistent. Same quality, a fraction of the time. #AI #Automation ``` It follows the before and after structure, stays under 150 words, and uses two hashtags. The 2-hour and 20-minute figures must be your own measured times. If you did not time it, do not post it. ## Test the file with five drafts before you rely on it A file can look complete and still fail. Test it before you use it for a week of posts. 1. Pick five topics that you posted about in the last three months. 2. Ask for one draft per topic. Name the pillar and the platform in each request. 3. Check each draft against every DO and DON'T line, and count the violations. 4. Put each draft beside the post that you published on that topic. 5. For each violation, find the rule that should have stopped it. 6. Rewrite that rule so that a reader can check it, and run the draft again. Here is a worked example. Three of five drafts end with "What do you think?" The file says "Ask at most one question per post", and each draft asks only one. The rule passed, but the ending is still generic. Add a new DON'T line: "End on a fact or a next step. Ask a question only when the post collects data." Run the three drafts again. If they now end on a fact, keep the rule. A second example: a draft uses "leverage", although the DON'T list bans that word. Search the top posts library for the word. If an old post uses it, the assistant copied it from the example. Edit the saved post, or write a note beside it that the word is not part of your voice now. ## Fix the three ways the file goes wrong The file fails in three predictable ways after a few months of use. The first failure is a file that grows too long. Each top post adds 150 words or more, and after 30 posts the examples outweigh the rules. The assistant then copies phrases from old posts. Keep 5 to 10 top posts in the file. Move older posts to an archive file that the assistant does not read. The second failure is a stale platform rule. You checked the link rule in January, and by June it may not hold. Each platform rule carries a date for this reason. Before you rely on a rule older than three months, check it against your own analytics. The third failure is two rules that conflict. The contrarian structure uses the set phrase "But here's what they're missing:". A voice rule that bans set phrases conflicts with it. The assistant then follows one rule on some drafts and the other rule on others. Write which rule wins inside the rule itself, for example "Set phrases are banned, except in the contrarian structure." ## Put the file where the assistant reads it The file helps only if the assistant reads it before each draft. In a chat app with projects, add `content.md` to the project that you use for social posts. Each new chat in that project can then read it. In Claude Code, keep `content.md` in the project folder, and add one line to `CLAUDE.md`: "Read `content.md` before you draft a post." Claude Code loads `CLAUDE.md` at the start of each session. For a template of that file, see [CLAUDE.md template examples](/articles/claude-md-template-examples). The file stays on your disk. The text that a session reads goes to the model provider for processing, and the retention terms depend on your plan. Keep client names and unpublished figures out of `content.md` unless your plan terms allow them. ## Start small and grow the file Follow this order: 1. Write 10 things you do and 10 things you avoid in your writing. 2. Add three or four content pillars, each with its angle. 3. Write one paragraph of rules for each platform you use. 4. Ask for drafts, and edit them. 5. After each post that performs, add it to the top posts library with the reason. Step 5 is where the file learns from your results. An edit you make on every draft is a missing voice rule; add it. [How corrections become part of the system](/articles/corrections-become-system-memory) shows how to make each fix reach the next draft. The assistant drafts; you publish. A post goes out under your name, so read every claim and every number first. [Who approves what an AI agent sends](/articles/who-approves-what-an-ai-agent-sends) covers how to set that rule when a team shares the account. ## Related: AI memory - [Measure AI first-draft editing time and cut it with persistent context](/articles/ai-first-draft-accuracy) - [Give AI a client file so email drafts start with the right context](/articles/ai-memory-before-and-after-email) - [Write an email context file so AI drafts sound like you](/articles/ai-memory-for-email-writing) ---- ```text .|########||. .|########||. .|########||. |##||. .||##|. |##||. .||##|. |##||. .||##|. |#|. .|#|. |#|. .|#|. |#|. .|#|. |#| |#| |#| |#| |#| |#| .#| |#. .#| |#. .#| |#. .#. .#| .#. .#| .#. .#| |#. .#| |#. .#| |#. .#| |# ...... #| |# ...... #| |# ...... #| .# ||#########| #| .# ||#########| #| .# ||#########| #| .#. |######||######|. .#. .#. |######||######|. .#. .#. |######||######|. .#. #. .##|###|##|#|######| .# #. .##|###|##|#|######| .# #. .##|###|##|#|######| .# || |##|#||||||||||||#||#| || || |##|#||||||||||||#||#| || || |##|#||||||||||||#||#| || # |#||||||||||||||||||||#| #. # |#||||||||||||||||||||#| #. # |#||||||||||||||||||||#| #. || |#||||||||||||||||||||||#| || || |#||||||||||||||||||||||#| || || |#||||||||||||||||||||||#| || # .#||||||||||||||||||||||||#| # # .#||||||||||||||||||||||||#| # # .#||||||||||||||||||||||||#| # || ....|||#||||||##|#|||#|#||##||||....|. || || ....|||#||||||##|#|||#|#||##||||....|. || || ....|||#||||||##|#|||#|#||##||||....|. || #. . ....|# ....#||| ||| .#||| ....#. .# #. . ....|# ....#||| ||| .#||| ....#. .# #. . ....|# ....#||| ||| .#||| ....#. .# # . ||||#| .#####|| . .#| .#||| ||||# #. # . ||||#| .#####|| . .#| .#||| ||||# #. # . ||||#| .#####|| . .#| .#||| ||||# #. .| ||||| #. |#|||||. || #. |#||. .||||| || .| ||||| #. |#|||||. || #. |#||. .||||| || .| ||||| #. |#|||||. || #. |#||. .||||| || || |.... #. ....|#. |. ...|. ....|| || || |.... #. ....|#. |. ...|. ....|| || || |.... #. ....|#. |. ...|. ....|| || #. .||||||##||||||##||####|||||||#||||||#| .# #. .||||||##||||||##||####|||||||#||||||#| .# #. .||||||##||||||##||####|||||||#||||||#| .# # .||####|###########################|. # # .||####|###########################|. # # .||####|###########################|. # # .#||||||.#.|| # |. ..# #| #|||||#| # # .#||||||.#.|| # |. ..# #| #|||||#| # # .#||||||.#.|| # |. ..# #| #|||||#| # # .#|||||| .. |# ## |#| ...#|#|||#| # # .#|||||| .. |# ## |#| ...#|#|||#| # # .#|||||| .. |# ## |#| ...#|#|||#| # # .#|||#|# .# .#| #| ##|.#..#|||||#| # # .#|||#|# .# .#| #| ##|.#..#|||||#| # # .#|||#|# .# .#| #| ##|.#..#|||||#| # # .#||||||############|######|||||#| # # .#||||||############|######|||||#| # # .#||||||############|######|||||#| # # |...||#||||||#||||#|#||||||#|||||#||| |.| # # |...||#||||||#||||#|#||||||#|||||#||| |.| # # |...||#||||||#||||#|#||||||#|||||#||| |.| # #. .. ||||# .|||##|. |#| .|| || .|||#. #| # .# #. .. ||||# .|||##|. |#| .|| || .|||#. #| # .# #. .. ||||# .|||##|. |#| .|| || .|||#. #| # .# || | ...|| ...##| | #| .|. |. ##### .. .| || || | ...|| ...##| | #| .|. |. ##### .. .| || || | ...|| ...##| | #| .|. |. ##### .. .| || || ||||| || ||||#| . |. |. |#. ||||| |#| || || || ||||| || ||||#| . |. |. |#. ||||| |#| || || || ||||| || ||||#| . |. |. |#. ||||| |#| || || .# |.....#|....|#.||||.|||##.|#|....||.#||.#. #. .# |.....#|....|#.||||.|||##.|#|....||.#||.#. #. .# |.....#|....|#.||||.|||##.|#|....||.#||.#. #. #.|||||||#############################| ||| .# #.|||||||#############################| ||| .# #.|||||||#############################| ||| .# || ##||#||#||||||#||#||#||#||##. || || ##||#||#||||||#||#||#||#||##. || || ##||#||#||||||#||#||#||#||##. || # .#|||||||||#||#|||||||||||#| # # .#|||||||||#||#|||||||||||#| # # .#|||||||||#||#|||||||||||#| # || |#||||||||||||||||||||||#| || || |#||||||||||||||||||||||#| || || |#||||||||||||||||||||||#| || .# |#||||||||||||||||||||#| #. .# |#||||||||||||||||||||#| #. .# |#||||||||||||||||||||#| #. || |#||||||||||||||||||#| || || |#||||||||||||||||||#| || || |#||||||||||||||||||#| || #. .######|#|#########| .# #. .######|#|#########| .# #. .######|#|#########| .# .#. |#####||||#####| .#. .#. |#####||||#####| .#. .#. |#####||||#####| .#. |# ||########|| #. |# ||########|| #. |# ||########|| #. |# ...... #| |# ...... #| |# ...... #| |#. .#| |#. .#| |#. .#| |#. .#. |#. .#. |#. .#. .#| |#. .#| |#. .#| |#. |#| |#| |#| |#| |#| |#| .|#|. .|#|. .|#|. .|#|. .|#|. .|#|. |##||. .||##| |##||. .||##| |##||. .||##| .||########||. .||########||. .||########||. Scale With Search 2026 [scalewithsearch.com](https://scalewithsearch.com) ```