\u23f1 9-minute read<\/p>\n
A content strategist in Ho Chi Minh City shared that she uses ChatGPT an average of 40 times a day. She opens a new conversation to write a brief for client A, switches to another tab to research the market for client B, then returns to the old tab to edit a blog post for project C.<\/p>\n
By the end of the day, she realized she had re-explained the project context at least 15 times to the same AI tool. Not because the AI was poor, but because her information organization forced each conversation to start from scratch.<\/p>\n
This is more common than many think. A 2023 Microsoft study shows that knowledge workers switch between work windows an average of over 1,200 times a day<\/strong>.<\/p>\n According to Gloria Mark at UC Irvine, each context switch costs an average of 23 minutes<\/strong> to regain focus.<\/p>\n When AI is added to the workflow, this problem doesn't just persist; it multiplies. Each new AI conversation is truly a complete reset. The tool doesn't know what you're doing, what you've done before, or where you want to go.<\/p>\n This article isn't about writing better prompts. It's about something far more important than prompts: context management<\/strong> \u2014 the art of organizing context so that AI (and your own brain) can operate at their best, instead of constantly circling through fragmented conversations.<\/p>\n Table of Contents<\/p>\n <\/p>\n Everyone has experienced this: You're writing an important email when the phone rings. You answer it for about 2 minutes, then return, but have to reread from the beginning to remember where you left off.<\/p>\n Cal Newport \u2014 author of Deep Work<\/em> and a computer science professor at Georgetown \u2014 calls this phenomenon attention residue<\/strong> (attention residue). When you switch from task A to task B, a part of your mind remains “stuck” on task A. The brain has no clean way to reset.<\/p>\n The real cost isn't in the switching time, but in the diminished quality of thought at both ends.<\/p>\n With AI, this phenomenon has an even more troublesome variant. Every time you open a new conversation, you have to re-explain what project you're working on, for whom, what the goal is, and what you've already tried.<\/p>\n You're not just losing time. You're rebuilding the entire context from scratch<\/strong>, both for the AI and for yourself.<\/p>\n Each time you rebuild, you lose subtle details accumulated from previous conversations. Specifically, this includes the phrasing AI has understood, the directions explored and discarded, and the implicit decisions no one had time to record.<\/p>\n For example, you're researching content strategy for the next quarter. The first conversation with AI analyzed 50 old blog posts, identified 3 content gaps, and proposed 12 new topics.<\/p>\n The second conversation, in another tab, is refining the brand voice. The third is writing an outline for the first article.<\/p>\n Three parallel conversations, three completely separate contexts. None knows what the others have done. The result is that you become the interpreter between your own AI versions<\/strong>.<\/p>\n Context tax = time spent re-explaining context + lost information between sessions + cognitive cost to synchronize in your head. For advanced AI users, this tax can account for 30\u201340% of total interaction time<\/strong> every day.<\/p>\n<\/div>\n Gerald Weinberg \u2014 author of Quality Software Management<\/em> \u2014 once measured something quite surprising. When you work on 2 projects simultaneously, time isn't split 50\/50 but is reduced to just 40% for each project<\/strong>, with 20% completely lost to context switching costs.<\/p>\n With 3 projects, each project gets only 20%. Nearly half the day's time evaporates into “remembering where I am.”<\/p>\n Therefore, if we apply these numbers to how you use AI \u2014 running 3 projects simultaneously on ChatGPT, for example \u2014 nearly half of your interaction time is spent reiterating context. Both for the AI, and for yourself.<\/p>\n <\/p>\n Every AI model has a limit called context window<\/strong> \u2014 l\u01b0\u1ee3ng th\u00f4ng tin t\u1ed1i \u0111a m\u00e0 m\u00f4 h\u00ecnh c\u00f3 th\u1ec3 “nh\u00ecn th\u1ea5y” trong m\u1ed9t l\u1ea7n t\u01b0\u01a1ng t\u00e1c. T\u00f9y m\u00f4 h\u00ecnh, con s\u1ed1 n\u00e0y dao \u0111\u1ed9ng t\u1eeb v\u00e0i ch\u1ee5c ngh\u00ecn \u0111\u1ebfn h\u00e0ng tri\u1ec7u t\u1eeb.<\/p>\n While it sounds expansive, in reality, the context window isn't the capacity you need to worry about. The real issue is how you organize the information within it<\/strong>.<\/p>\n The easiest way to visualize a context window is to think of a physical desk. You might have a 2-meter wide desk, but if it's cluttered with unsorted papers, you still won't find the document you need in 10 seconds.<\/p>\n John Sweller's research on cognitive load shows that human working memory can only effectively process about 4\u20135 chunks of information<\/strong> at once.<\/p>\n AI operates on a similar logic. No matter how large the context window, output quality significantly decreases when input information lacks clear structure.<\/p>\n So, the most effective AI users are often not those who write the longest prompts. They are the ones who design the best context<\/strong>. They know how to pre-determine what information to include, in what order, and at what level of detail.<\/p>\n In the AI community, this approach is being called context engineering<\/strong>. It emphasizes providing the right context rather than refining the wording in the prompt.<\/p>\n “The hottest new programming skill is not prompting, it’s context engineering \u2014 the art of providing all the right context at the right time to the model.”<\/p>\n \u2014 Andrej Karpathy, former Director of AI at Tesla<\/p>\n<\/div>\n Princeton psychologist George Miller demonstrated in his classic study “The Magical Number Seven” that humans process information most effectively when it is grouped into meaningful units<\/strong>. This principle applies very directly to how you should organize context for AI.<\/p>\n For example, you need AI to write a pitch email for a new client. What most people do is paste the entire 5-page brief into the prompt, along with the communication history and the price list.<\/p>\n They hope AI will figure out the core message. The result is often a long, rambling email that doesn't know who it's talking to.<\/p>\n However, if you only send 3 lines \u2014 the goal is to upsell the premium package, the target audience is a CFO interested in ROI, and the tone is professional and concise \u2014 then AI returns a focused email on the first try. Less information, but the right information. That's context engineering in everyday practice.<\/p>\n <\/p>\n Tiago Forte \u2014 author of Building a Second Brain<\/em> \u2014 has a rather apt term for it. This personal knowledge management method is being adopted by over 500,000 people.<\/p>\n \u00d4ng kh\u00f4ng d\u00f9ng t\u1eeb “kho l\u01b0u tr\u1eef” hay “h\u1ec7 th\u1ed1ng ghi ch\u00fa”. Thay v\u00e0o \u0111\u00f3, \u00f4ng g\u1ecdi second brain l\u00e0 structured external memory<\/strong> \u2014 meaning a system outside your brain that helps you store, organize, and retrieve information without relying on your memory.<\/p>\n Before the age of AI, the second brain primarily served a single purpose: to help you recall what you've read, thought, and learned.<\/p>\n However, in the age of AI, it gains an additional function that few immediately realize: providing context for AI<\/strong>. When you have a well-organized personal knowledge system, you no longer need to re-explain the context every time you start a new conversation.<\/p>\n You just need to point the AI to the right document, the right note, the right recorded decision. As a result, the AI can pick up where you left off.<\/p>\n There's a very useful principle from Niklas Luhmann's Zettelkasten system that you should apply. Each note should contain only a single idea<\/strong> and have clear links<\/strong> to related ideas.<\/p>\n This is precisely the structure that AI processes best. Instead of sifting through a 50-page document, AI only needs to review 5 concise notes. Each note covers a single topic, with clear links to each other.<\/p>\n K\u1ebft qu\u1ea3 l\u00e0 AI “hi\u1ec3u” nhanh h\u01a1n. V\u00e0 b\u1ea1n c\u0169ng ki\u1ec3m so\u00e1t \u0111\u01b0\u1ee3c ch\u00ednh x\u00e1c AI \u0111ang nh\u00ecn th\u1ea5y g\u00ec.<\/p>\n PARA + AI \u2014 four layers of context:<\/strong><\/p>\n Projects<\/strong> \u2014 ongoing projects: the primary context you feed to AI daily (brief, timeline, deliverables).<\/p>\n Areas<\/strong> \u2014 areas of responsibility: the foundational context AI needs to understand your role (brand guidelines, team processes, standards).<\/p>\n Resources<\/strong> \u2014 reference resources: additional context for in-depth exploration (research, case studies, frameworks).<\/p>\n Archive<\/strong> \u2014 archived: historical AI context is not visible unless actively retrieved.<\/p>\n<\/div>\n David Allen \u2014 author of Getting Things Done<\/em> \u2014 once said that “Your brain is for having ideas, not for holding them.”<\/strong>.<\/p>\n In the age of AI, this principle holds even truer in a way Allen couldn't have imagined. Your brain is also not the place to store context for AI.<\/p>\n If you keep trying to remember “what we discussed in yesterday’s conversation” \u2014 that’s precisely when you need an external system to do that work for you.<\/p>\n <\/p>\n When working with AI on multiple parallel projects \u2014 content writing, market research, strategy development \u2014 a three-tiered structure significantly reduces the “restart” time for each session.<\/p>\n This isn't a theoretical framework. This is an organizational method adopted by many content teams and freelancers after encountering the exact problem of “re-explaining context every day” described in the beginning of this article.<\/p>\n This is a single, concise file (under 500 words) containing everything the AI needs to know to start working with you on a specific project. It includes project goals, target audience, tone of voice, decisions made, and key constraints.<\/p>\n Each time you start a new conversation, send this context file first. The AI goes from zero to 80% understanding in 10 seconds, instead of you spending 5 minutes explaining from scratch.<\/p>\n Specifically, when creating content for a brand, the context file includes brand positioning, 5 main keywords, target persona, a list of published content, and a summarized style guide.<\/p>\n This is a Notion, Obsidian, or Google Drive system containing more detailed documents, organized using the PARA or Zettelkasten method. The AI doesn't need to \"see\" the entire knowledge base. You only extract relevant sections when needed.<\/p>\n For example, a content strategist might have a \"Brand Assets\" folder containing 30 documents. But when asking the AI to write a blog post, they only need to pull out the 3 most relevant documents: editorial guidelines, keyword research for that topic, and the approved outline.<\/p>\n This is something few people think about but makes the biggest difference. You record somewhere (a short note, a comment in a file) what was completed in the previous session<\/em> and what the next steps are<\/em>.<\/p>\n When you return to the project after 2 days, instead of reopening an old conversation or starting a new one, you send the AI: context file (Layer 1) + session state (Layer 3).<\/p>\n The AI immediately knows where you are, what has been done, and how to proceed.<\/p>\n Practical example \u2014 3-layer workflow:<\/strong><\/p>\n You are working on the Q2 content strategy for an edtech startup.<\/p>\n Monday Morning:<\/strong> Open Claude \u2192 send context file “Q2-content-strategy.md” \u2192 request content gap analysis \u2192 AI outputs 12 topics. You record the session state: “12 topics generated, need to validate with GA4 data”.<\/p>\n Tuesday Morning:<\/strong> Open Claude \u2192 send context file + session state + GA4 data table (extracted from knowledge base) \u2192 AI filters down to 7 topics with search volume. Session state updates: “7 topics validated, need to write outlines for the first 3 articles”.<\/p>\n Wednesday Morning:<\/strong> Context file + session state \u2192 AI writes 3 outlines, knowing exactly which step it's on, without asking any follow-up questions. Total context reconstruction time: under 30 seconds per day.<\/p>\n<\/div>\n This architecture simultaneously solves three problems. Each session starts faster because there's no need to re-explain. AI receives the necessary information instead of being overwhelmed by superfluous details. And you don't lose your workflow momentum between days.<\/p>\n <\/p>\n Most people open ChatGPT and start with “I'm working on a project about\u2026” then type 3\u20134 paragraphs of explanation. They repeat this every time, every conversation, every day.<\/p>\n There's a much simpler way. Write a context file of about 300 words just once, then send it at the beginning of each conversation.<\/p>\n AI will understand about 80% of the context in just the first 10 seconds. You won't need to spend 5 minutes re-explaining each time. For a 4-week project, an initial 15-minute investment can yield 3\u20135 hours in return.<\/p>\n Context file template \u2014 5 lines are enough:<\/strong><\/p>\n Project:<\/strong> Q2 Content Strategy \u2014 Edtech Startup XYZ Most people end their AI sessions by closing the tab. Tomorrow, they reopen it and scroll back through 50 messages to find “where they left off yesterday”. Or, more simply, they start a new conversation and explain everything from scratch.<\/p>\n Additionally, there's a much better way. Before closing the conversation, ask the AI to summarize in 3\u20135 lines: what was done, what decisions were made, and what the next steps are.<\/p>\n Copy that summary into your notes, and tomorrow, send it back to the AI along with your context file. This takes only 30 seconds but creates continuity between sessions \u2014 something most current AI workflows completely lack.<\/p>\n Many people have a habit of starting a new conversation for each question, even when all questions belong to the same project. Each new conversation means the AI completely forgets everything prior.<\/p>\n If you maintain a continuous conversation, context accumulates with each interaction. The AI understands you more deeply after every exchange.<\/p>\n Moreover, only start a new conversation when the AI begins to show signs of “forgetting” the earlier parts of the conversation. At that point, a context file combined with the previous session's summary will help the new conversation catch up very quickly.<\/p>\n A very common mistake is simultaneously brainstorming 10 ideas, writing an article, and editing \u2014 all in the same conversation. The result is that the AI confuses discarded ideas with chosen directions. This leads to self-contradictory output.<\/p>\n Cal Newport in his book Deep Work<\/em> offers a highly applicable principle here. Clearly separate one session solely for research and decision-making, then use another session solely for execution based on that decision.<\/p>\n When separated this way, the context window isn't diluted by outdated exploratory content. Consequently, the AI can focus precisely on the assigned task.<\/p>\n <\/p>\n \ud83d\udccc Key takeaways<\/p>\n <\/p>\n <\/p>\n Start with a single project. Write a 300-word context file for that project, including goals, audience, tone, key decisions, and constraints.<\/p>\n Use this context file to initiate every AI conversation. End each session with a 3\u20135 line session summary. After 1 week, you will clearly see the difference \u2014 and naturally want to expand to other projects.<\/p>\n Notion is suitable for teams due to its good collaboration features and flexible database structure. Obsidian is suitable for individuals who want data control, as it supports linking between notes and local storage.<\/p>\n Additionally, Google Docs\/Drive is suitable when simplicity and quick sharing are needed. The most important thing is not which tool \u2014 but having clearly structured information that can be quickly extracted when needed for AI.<\/p>\n Khi AI b\u1eaft \u0111\u1ea7u “qu\u00ean” nh\u1eefng g\u00ec b\u1ea1n \u0111\u00e3 th\u1ea3o lu\u1eadn \u1edf \u0111\u1ea7u cu\u1ed9c h\u1ed9i tho\u1ea1i \u2014 \u0111\u00f3 l\u00e0 d\u1ea5u hi\u1ec7u r\u00f5 r\u00e0ng nh\u1ea5t. C\u1ee5 th\u1ec3 l\u00e0 AI h\u1ecfi l\u1ea1i th\u00f4ng tin b\u1ea1n \u0111\u00e3 cung c\u1ea5p, \u0111\u01b0a ra g\u1ee3i \u00fd m\u00e2u thu\u1eabn v\u1edbi quy\u1ebft \u0111\u1ecbnh tr\u01b0\u1edbc \u0111\u00f3, ho\u1eb7c b\u1ecf qua r\u00e0ng bu\u1ed9c b\u1ea1n \u0111\u00e3 n\u00eau r\u00f5.<\/p>\n At that point, start a new conversation with a context file + session summary from the previous session. AI will catch up in a few seconds.<\/p>\n <\/p>\n The 3-layer architecture above \u2014 context file, knowledge base, session state \u2014 requires a platform for implementation. You can build it from scratch on Notion or Obsidian, or start with a pre-designed structure.<\/p>\n Practical Tools<\/p>\n Practical Knowledge Base<\/strong> \u2014 a Notion template designed using the PARA method. A ready-made structure is available for both personal and team knowledge management, optimized for context extraction when working with AI.<\/p>\n See details \u2192<\/a><\/p>\n<\/div>\n Context switching costs you 23 minutes every time you switch tasks. With AI, the problem multiplies \u2014 each new conversation is a complete reset. This article guides you on building a context management system to work more effectively with AI.<\/p>","protected":false},"author":10,"featured_media":17485,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_yoast_wpseo_focuskw":"context management","_yoast_wpseo_title":"","_yoast_wpseo_metadesc":"Context management quy\u1ebft \u0111\u1ecbnh ch\u1ea5t l\u01b0\u1ee3ng output c\u1ee7a AI h\u01a1n c\u1ea3 prompt. H\u1ecdc c\u00e1ch t\u1ed5 ch\u1ee9c ng\u1eef c\u1ea3nh \u0111\u1ec3 d\u00f9ng AI hi\u1ec7u qu\u1ea3 g\u1ea5p 3 l\u1ea7n.","_yoast_wpseo_linkdex":"81","_yoast_wpseo_content_score":"75","footnotes":"","_featured_image_en":17485},"categories":[16,17],"tags":[127,136,66],"class_list":["post-16847","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-blog","category-business","tag-ai-van-hanh","tag-context-management","tag-second-brain"],"yoast_head":"\n\n
\nEvery time you re-explain context to an AI, you're paying an invisible tax.<\/h2>\n
AI and the Cost of Rebuilding Context<\/h3>\n
The actual numbers on context switching costs<\/h3>\n
\nThe context window isn't a technical limitation \u2014 it's an architectural mindset.<\/h2>\n
The Desk and Cognitive Load<\/h3>\n
George Miller's Principle of Information Chunking<\/h3>\n
\nA second brain isn't a storage vault \u2014 it's a structured external memory.<\/h2>\n
Second brain in the age of AI<\/h3>\n
Zettelkasten Principle: one idea, one note<\/h3>\n
\nContext architecture: three layers every knowledge worker should design.<\/h2>\n
Tier 1: Context file \u2014 project identity document<\/h3>\n
Layer 2: Knowledge base \u2014 categorized resource library<\/h3>\n
Layer 3: Session state \u2014 working session status<\/h3>\n
\nFour habits that will transform how you work with AI.<\/h2>\n
Habit 1: Write a context file before starting a project<\/h3>\n
\nGoal:<\/strong> Increase organic traffic by 40%, focusing on the keyword cluster “learn English online”
\nAudience:<\/strong> Working professionals aged 25\u201335, eager to learn but time-constrained, accustomed to using apps.
\nTone:<\/strong> Professional yet approachable, non-academic, using everyday examples.
\nDecision Made:<\/strong> Not writing about IELTS (saturated market), focusing on business English + communication.
\nConstraints:<\/strong> 4 posts\/month, 1200\u20131500 words each, published Wednesdays at 8 AM.<\/p>\n<\/div>\nHabit 2: End each AI session with a brief summary.<\/h3>\n
Habit 3: Maintain a continuous conversation for each project<\/h3>\n
Habit 4: Separate Thinking Sessions from Execution Sessions<\/h3>\n
\n\n
\n
\nFAQ \u2014 Frequently Asked Questions<\/h2>\n
Where to start building a context management system?<\/h3>\n
Which tools are suitable for building a knowledge base for AI workflows?<\/h3>\n
How do you know when the context window is getting full and it's time to start a new conversation?<\/h3>\n
\nRelated Posts<\/h2>\n
\n