--- title: "Create an AI memory file in ten steps and load it every session" description: "Write a CLAUDE.md with identity, voice, projects, domain context, preferences, and templates. Save it where your assistant reads it and keep it current." canonical: "https://scalewithsearch.com/articles/how-to-create-ai-memory-file" 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) # Create an AI memory file in ten steps and load it every session. You explain yourself to ChatGPT, Claude, and Gemini every time you open a new chat. You want the assistant to know who you are, what you do, and how you work. The answer is not a memory feature. It is a file: one Markdown file with structured sections that holds what the assistant needs to know. You write it once and maintain it. This article builds the file in ten steps. The examples use a synthetic marketing consultant. Replace every placeholder with your own facts. ## Know what an AI memory file is An AI memory file is a plain text document, usually in Markdown, that holds your context. It works like a user manual for how you work. When an assistant reads it, the assistant knows five things: - who you are: role, industry, and business model; - what you work on: active projects, clients, and deadlines; - how you communicate: tone, style, and banned phrases; - your workflows: standard processes, templates, and SOPs; - domain knowledge: your clients, tools, and team structure. You stop typing "I am a real estate agent in Miami, I focus on luxury condos, I prefer direct communication" at the start of each conversation. The assistant reads it from the file. ## Use Markdown Markdown is plain text with simple formatting. People and assistants can both read it, and it works in every editor. You can write it in Notepad on Windows, TextEdit on Mac, VS Code, or Obsidian, which suits larger knowledge systems. It needs no special software and no proprietary format. [Why plain text is the durable layer for AI memory](/articles/plain-text-ai-memory) explains why that matters when you change tools. ## Step 1: create the file Open your text editor and create a new file. Name it `CLAUDE.md`, `AI-CONTEXT.md`, or another clear name. Claude Code looks for `CLAUDE.md` by name, so use that name if you use Claude Code. The `.md` extension marks the file as Markdown. If your editor does not support Markdown, save it as `.txt`. The content still works. ## Step 2: start with identity The first section says who you are. Be specific, because a generic description produces generic output. A weak identity line: > I am a consultant who helps businesses with marketing. A useful identity line: > I am a fractional CMO for B2B SaaS companies. I help companies move from founder-led sales to repeatable marketing systems. I build content engines, tune paid acquisition, and set up attribution tracking. A typical engagement runs 6 to 12 months. The second version tells the assistant your market, your service model, and your engagement length. That changes how it drafts proposals, writes email, and suggests strategy. Keep your prices in a separate rate file, so that one change updates every draft. Use this structure: ```markdown ## Identity I am [name], [role] at [company or industry]. I [what you do, specifically]. Focus: [target market or niche] Business model: [how you earn revenue] Experience: [years and relevant background] ``` ## Step 3: define your voice This section controls tone, structure, and style. ```markdown ## Voice Tone: direct, conversational, no corporate jargon. Use contractions. Sound like a person, not a press release. Structure: lead with the main point, then the details. No long intros. Format: - Bullet lists for lists of items - Short sentences, under 25 words where possible - Active voice - Specific numbers, not vague claims Banned words and phrases: - solutions, leverage, synergy, ecosystem - "moving forward," "circle back," "touch base" - "cutting-edge," "world-class," "innovative" Style: - Use "you" and "I," not "one," and "we" only for a team - End a sentence when the idea is complete - If a line sounds like marketing copy, rewrite it ``` Concrete rules produce concrete results. The assistant has an explicit list to check each draft against. If a banned phrase still appears, point to the rule and make it more specific. ## Step 4: list your projects List active clients, internal work, launches, and campaigns. ```markdown ## Active projects Client A (CRM platform for real estate) - Engagement start: Jan 2026 - Current phase: content strategy and SEO audit - Deliverable due: Feb 28 (topical authority map and Q1 content calendar) - Primary contact: VP Marketing Client B (supply chain software) - Engagement start: Nov 2025 - Current phase: paid acquisition work - Focus: Google Ads, LinkedIn, attribution setup - Primary contact: founder and CEO Internal: "AI for marketers" course - Launch date: March 15, 2026 - Status: module 3 of 6 complete - Platform: Teachable ``` When you ask for a status email to Client A, the assistant knows the phase, the deadline, and the contact. ## Step 5: add domain context Record the knowledge the assistant needs about your work: clients, tools, workflows, and team structure. ```markdown ## Clients Client A - Product: CRM for real estate agents - Stage: Series A, 40 employees - My role: content and SEO strategy - Contact details: in the CRM, not in this file Client B - Product: supply chain visibility platform - Stage: seed, 12 employees - My role: paid acquisition and attribution ## Tools - Project management: Notion - Communication: Slack for clients, email for prospects - Analytics: Google Analytics 4, Looker Studio - SEO: Ahrefs, Screaming Frog - Paid ads: Google Ads, LinkedIn Campaign Manager ## Standard workflows New client onboarding: 1. Kickoff call (45 minutes): goals, timeline, access 2. Audit (weeks 1 to 2): current state analysis 3. Strategy document (week 3): recommendations and roadmap 4. Execution (week 4 onward) Monthly client report: - Sent on the first Monday of each month - Sections: metrics, wins, challenges, next month's plan - Format: Google Doc, exported to PDF - Always include 3-month trend graphs ``` With this detail, the assistant drafts a client report in the correct format, sections, and timing without instructions from you. ## Step 6: record your preferences Preferences are the small details you correct after every draft. ```markdown ## Preferences Dates: YYYY-MM-DD format (2026-01-28, not Jan 28, 2026) Names: give full context at first mention, such as "the VP Product at Client A," not a first name alone Email: - Subject lines under 50 characters, no clickbait - Greeting: "Hi [name]," - Closing: "Thanks," or "Best," - Always end with a clear next step Meeting notes: - Date and attendees at the top - Decisions as a bullet list - Action items with who, what, and when - Open questions - Next meeting date ``` Write each preference once, and stop correcting it. ## Step 7: include templates If you send similar email or documents often, add templates. ```markdown ## Email templates ### New prospect outreach Subject: [specific problem]: quick question Hi [name], I saw [specific detail about the company or its content]. [One sentence that connects that detail to a problem you solve.] I work with [type of company] to [outcome you deliver]. Worth a 15-minute call to see if there is a fit? [Calendar link] Thanks, [Your name] ### Client status update Subject: [client name] update, [month YYYY] Hi [name], Here is what we shipped this month. Completed: - [item] In progress: - [item] Next month: - [item] Questions are welcome. Otherwise, we keep moving. Thanks, [Your name] ``` When you ask for an outreach email or a status update, the assistant starts from your template. ## Step 8: save the file where the assistant can read it You have three choices: 1. Project folder. If you use Claude Code or another assistant that reads local files, save `CLAUDE.md` in your main working folder. Claude Code reads it automatically. 2. Obsidian vault. If you plan a larger system, save the file in a vault. You can link it to other notes and grow a multi-file system. The [vault organization guide](/articles/how-to-organize-ai-knowledge-vault) shows the structure. 3. Cloud storage. Save the file in Google Drive, Dropbox, or OneDrive to reach it from any device. Make sure you can download it when you need it. ## Step 9: load the file in each tool How you load the file depends on the tool. This list reflects a January 2026 observation: | Tool | How the file loads | |---|---| | Claude Code | Reads `CLAUDE.md` from the working folder at the start of each session, with no manual step | | ChatGPT | Upload the file at the start of each session; it applies for that conversation | | Claude Projects | Upload the file to a project once; it applies to every chat in that project | | Any assistant | Paste the content at the start of an important conversation | ChatGPT custom instructions can hold a short version. Each field held about 1,500 characters in January 2026, which is roughly one paragraph. The [ChatGPT instruction limits article](/articles/chatgpt-custom-vs-project-instructions-limits) tracks the published figures. A memory file has no field limit. Its practical limit is the context window of the model that reads it. ## Step 10: update the file regularly The file is not static. Update it as your work changes: - New client: add it to the clients section. - Finished project: remove it from active projects. - Changed preference: edit the voice or preferences section. - New workflow: document it. Set a monthly reminder to review the file. The review takes about 10 minutes and keeps the assistant accurate. ## Test the file in a fresh session A file that exists is not yet a file that works. Test it before you rely on it: 1. Open a new session with no other context. 2. Ask, "Who am I, and what am I working on this week?" 3. Ask for a status email to Client A. 4. Ask for a LinkedIn post about a finished project. 5. Ask about a client that is not in the file. Score each answer. Step 2 must name your role and the active projects from the file. Step 3 must use the current phase, the deadline, and your greeting. Step 4 must contain no banned phrase. Step 5 must say that the file has no record of that client. An invented client history is a failure. When an answer fails, find the line that should have prevented it. If the line is missing, add it. If the line exists and the assistant ignored it, make the line more specific. "Short sentences" is weaker than "Keep each sentence under 20 words." Run the test again after each change, and keep the dated results. They show which sections change the output and which ones the assistant does not use. ## Start from a minimal file A starter file can be short: ```markdown # AI context ## Identity I am a solo SEO consultant for service businesses. My clients are practices and small teams under 10 people who want to rank for high-intent keywords and use AI to produce content without losing quality. ## Voice - Tone: direct, conversational, no corporate jargon - Structure: lead with the main point - Format: bullet lists, short sentences, active voice - Banned words: solutions, leverage, ecosystem, cutting-edge ## Active projects - Client onboarding system redesign (due Feb 2026) - SEO course module 3 (launch March 2026) - Weekly newsletter (published Thursdays) ## Preferences - Dates: YYYY-MM-DD - Email greeting: "Hi [name]," - Always include next steps in email ``` That file is about 100 words. It is enough to improve every conversation. Expand it from there: add clients, document workflows, and build toward the full structure. The [CLAUDE.md business template](/articles/claude-md-template-business-context) shows a larger finished version. ## Avoid four common mistakes Vague identity: "I am a consultant" does not help. "I am a fractional CFO for SaaS companies that prepare for Series A fundraising" does. Outdated information: if the file lists a project that ended three months ago, the assistant refers to it. Review the file each month. Too much at once: you do not need 50 pages on day one. Start with 300 words and expand when a gap appears. No use: the file helps only when the assistant reads it. If you still open ChatGPT without loading your context, you get none of the benefit. Also keep secrets out. The file stays on your disk, but its content goes to the model provider when a session reads it. Do not write passwords, API keys, or sensitive client records into it. ## Related: AI memory - [Build an AI second brain in one 90-minute session](/articles/how-to-build-ai-second-brain) - [Close the AI trust gap by giving each role a memory layer](/articles/stanford-ai-index-role-memory-gap) - [Install Obsidian and Claude Code, then connect your first AI vault](/articles/how-to-set-up-ai-vault) ---- ```text .|########||. .|########||. .|########||. |##||. .||##|. |##||. .||##|. |##||. .||##|. |#|. .|#|. |#|. .|#|. |#|. .|#|. |#| |#| |#| |#| |#| |#| .#| |#. .#| |#. .#| |#. .#. .#| .#. .#| .#. .#| |#. .#| |#. .#| |#. .#| |# ...... #| |# ...... #| |# ...... #| .# ||#########| #| .# ||#########| #| .# ||#########| #| .#. |######||######|. .#. .#. |######||######|. .#. .#. |######||######|. .#. #. .##|###|##|#|######| .# #. .##|###|##|#|######| .# #. .##|###|##|#|######| .# || |##|#||||||||||||#||#| || || |##|#||||||||||||#||#| || || |##|#||||||||||||#||#| || # |#||||||||||||||||||||#| #. # |#||||||||||||||||||||#| #. # |#||||||||||||||||||||#| #. || |#||||||||||||||||||||||#| || || |#||||||||||||||||||||||#| || || |#||||||||||||||||||||||#| || # .#||||||||||||||||||||||||#| # # .#||||||||||||||||||||||||#| # # .#||||||||||||||||||||||||#| # || ....|||#||||||##|#|||#|#||##||||....|. || || ....|||#||||||##|#|||#|#||##||||....|. || || ....|||#||||||##|#|||#|#||##||||....|. || #. . ....|# ....#||| ||| .#||| ....#. .# #. . ....|# ....#||| ||| .#||| ....#. .# #. . ....|# ....#||| ||| .#||| ....#. .# # . ||||#| .#####|| . .#| .#||| ||||# #. # . ||||#| .#####|| . .#| .#||| ||||# #. # . ||||#| .#####|| . .#| .#||| ||||# #. .| ||||| #. |#|||||. || #. |#||. .||||| || .| ||||| #. |#|||||. || #. |#||. .||||| || .| ||||| #. |#|||||. || #. |#||. .||||| || || |.... #. ....|#. |. ...|. ....|| || || |.... #. ....|#. |. ...|. ....|| || || |.... #. ....|#. |. ...|. ....|| || #. .||||||##||||||##||####|||||||#||||||#| .# #. .||||||##||||||##||####|||||||#||||||#| .# #. .||||||##||||||##||####|||||||#||||||#| .# # .||####|###########################|. # # .||####|###########################|. # # .||####|###########################|. # # .#||||||.#.|| # |. ..# #| #|||||#| # # .#||||||.#.|| # |. ..# #| #|||||#| # # .#||||||.#.|| # |. ..# #| #|||||#| # # .#|||||| .. |# ## |#| ...#|#|||#| # # .#|||||| .. |# ## |#| ...#|#|||#| # # .#|||||| .. |# ## |#| ...#|#|||#| # # .#|||#|# .# .#| #| ##|.#..#|||||#| # # .#|||#|# .# .#| #| ##|.#..#|||||#| # # .#|||#|# .# .#| #| ##|.#..#|||||#| # # .#||||||############|######|||||#| # # .#||||||############|######|||||#| # # .#||||||############|######|||||#| # # |...||#||||||#||||#|#||||||#|||||#||| |.| # # |...||#||||||#||||#|#||||||#|||||#||| |.| # # |...||#||||||#||||#|#||||||#|||||#||| |.| # #. .. ||||# .|||##|. |#| .|| || .|||#. #| # .# #. .. ||||# .|||##|. |#| .|| || .|||#. #| # .# #. .. ||||# .|||##|. |#| .|| || .|||#. #| # .# || | ...|| ...##| | #| .|. |. ##### .. .| || || | ...|| ...##| | #| .|. |. ##### .. .| || || | ...|| ...##| | #| .|. |. ##### .. .| || || ||||| || ||||#| . |. |. |#. ||||| |#| || || || ||||| || ||||#| . |. |. |#. ||||| |#| || || || ||||| || ||||#| . |. |. |#. ||||| |#| || || .# |.....#|....|#.||||.|||##.|#|....||.#||.#. #. .# |.....#|....|#.||||.|||##.|#|....||.#||.#. #. .# |.....#|....|#.||||.|||##.|#|....||.#||.#. #. #.|||||||#############################| ||| .# #.|||||||#############################| ||| .# #.|||||||#############################| ||| .# || ##||#||#||||||#||#||#||#||##. || || ##||#||#||||||#||#||#||#||##. || || ##||#||#||||||#||#||#||#||##. || # .#|||||||||#||#|||||||||||#| # # .#|||||||||#||#|||||||||||#| # # .#|||||||||#||#|||||||||||#| # || |#||||||||||||||||||||||#| || || |#||||||||||||||||||||||#| || || |#||||||||||||||||||||||#| || .# |#||||||||||||||||||||#| #. .# |#||||||||||||||||||||#| #. .# |#||||||||||||||||||||#| #. || |#||||||||||||||||||#| || || |#||||||||||||||||||#| || || |#||||||||||||||||||#| || #. .######|#|#########| .# #. .######|#|#########| .# #. .######|#|#########| .# .#. |#####||||#####| .#. .#. |#####||||#####| .#. .#. |#####||||#####| .#. |# ||########|| #. |# ||########|| #. |# ||########|| #. |# ...... #| |# ...... #| |# ...... #| |#. .#| |#. .#| |#. .#| |#. .#. |#. .#. |#. .#. .#| |#. .#| |#. .#| |#. |#| |#| |#| |#| |#| |#| .|#|. .|#|. .|#|. .|#|. .|#|. .|#|. |##||. .||##| |##||. .||##| |##||. .||##| .||########||. .||########||. .||########||. Scale With Search 2026 [scalewithsearch.com](https://scalewithsearch.com) ```