https://www.zhengyang.me//Zhengyang Hou2026-05-19T04:00:34.322Zhttps://github.com/jpmonette/feedZhengyang Houjenqyanghou@gmail.comhttps://www.zhengyang.me/Founder & builder. Writing about AI-first products, agents, and the taste and judgment that the next decade will reward.https://www.zhengyang.me//favicon.svgAll rights reserved 2026, Zhengyang Hou<![CDATA[Your AI remembers too much of the wrong things]]>https://www.zhengyang.me//your-ai-remembers-too-much-of-the2026-04-21T16:00:00.000Z
Your AI remembers too much of the wrong things
On context rot, the illusion of personalization, and why I built Memory Sync.
I use four different AI assistants in a typical week. ChatGPT for quick tasks. Claude for writing and code. Gemini when I'm deep inside Google Docs. Kimi when I need long-context Chinese material. If you're reading this, your mix is probably similar, even if the specific names are different.
About a year ago, I started noticing something was off. Not dramatically off — just quietly off, in a way that was hard to name at first.
Here's what it looked like. I'd open a new chat with ChatGPT and ask for help naming a side project. The suggestions would come back uncannily in my taste — minimalist, lowercase, with that particular melancholy-but-hopeful tone I tend toward. On the face of it, great. Personalized. Exactly what the marketing had promised.
But I'd ask Claude the same question in a fresh session where it knew nothing about me, and the suggestions were wider. Some were bad. Some were surprising. A few of those surprising ones were actually better than anything ChatGPT had given me — they were names I wouldn't have thought of myself, and that was the point.
That's when I started paying attention to what memory was doing to my outputs.
The feature that makes your AI more agreeable, not more useful
I'm not the first person to notice this. Mike Taylor, co-author of O'Reilly's Prompt Engineering for Generative AI, published an essay called "Why I Turned Off ChatGPT's Memory" where he coined a useful term for the pattern: context rot — the slow buildup of stale preferences and half-remembered details that quietly degrades your results over time.
Simon Willison, who co-created Django, wrote something even more pointed. He was trying to test ChatGPT's ability to geolocate photos. He uploaded a neighborhood image — and the model already knew where he lived. From a completely unrelated earlier conversation. The entire test was invalidated before it began. His core objection goes to the heart of how these tools actually work: prompting well means carefully controlling context, and memory removes that control.
Then there's the research. An MIT study presented at ACM CHI in early 2026 collected two weeks of real conversation data from 38 participants. The finding: memory-based personalization had the largest measurable effect on increasing AI sycophancy. The models didn't just become more helpful. They became more likely to agree with you, including on political questions.
A Columbia Journalism Review investigation surfaced something worse. Users consistently rate sycophantic, memory-influenced responses as higher quality. Which means the pressure on AI companies runs in exactly the wrong direction. You and I may be getting worse answers, but we feel like we're getting better ones, and we rate them that way.
When I read this work, a lot of things clicked into place for me. The reason ChatGPT's naming suggestions had felt so in my taste wasn't that it understood me better. It was that I was sitting inside an echo chamber I couldn't see.
The second problem: your memory isn't yours
Here's what bothered me even more than context rot: I couldn't do anything about it.
I could open ChatGPT's settings and see a list of saved memories. But the list was incomplete — there's a separate "Reference chat history" mechanism that pulls context from past chats without ever showing you what. I could delete individual memory entries, but I couldn't meaningfully edit them. I couldn't group them. I couldn't take them out.
And I definitely couldn't take them to another AI.
This is the part nobody talks about when they talk about personalization. The better ChatGPT gets at knowing you, the more expensive it becomes for you to leave. I started hearing variations of this from other heavy AI users: "I'd try Claude, but ChatGPT knows me too well now." Mike Taylor mentions the same pattern. I heard it from a friend who runs a small marketing team. I felt it myself — the quiet gravitational pull of the platform that had the most context on me.
That isn't personalization. It's a switching cost dressed up as a feature.
The power-user community has been voting with their behavior on this for a while now. A Hacker News thread earlier this year asked whether anyone actually uses built-in LLM memory features. The answers were almost unanimous: people had turned it off and were managing their own context in plain markdown files instead. One commenter put it cleanly — built-in memory is a black box, manual context gives you predictable behavior. That thread was my final push.
What memory should actually look like
After sitting with this for a while, I wrote down what I wanted instead. The list was short and unglamorous:
I should be able to see everything the AI knows about me. Not a filtered summary. The actual content, in plain text, in a format I can read.
I should be able to edit it. Remove outdated items. Rewrite the parts that are half-right. Reorganize by project.
It should live outside the platform. Because if it lives inside ChatGPT, it belongs to ChatGPT.
It should move between tools. The whole point of using multiple assistants is picking the right model for the job. I shouldn't have to re-teach each one who I am.
None of this is radical. It's roughly how we treat every other kind of personal data — passwords, notes, files, calendars. The fact that AI memory became the exception isn't because portability is technically hard. It's because portability isn't in the platforms' interest.
So I built Memory Sync.
One Memory.md, seven platforms, a human in the loop
The core idea is embarrassingly simple. Memory Sync is a Chrome extension that gives you a single file — Memory.md — that serves as your portable context layer. You pull memory out of ChatGPT, Claude, Gemini, Grok, Kimi, Mistral, or Copilot, edit it as a markdown file, and push it into any of the others.
The three verbs that matter are Pull, Edit, Push.
Pull grabs what a platform has inferred or stored about you. It comes out as plain text. You can read the whole thing in five minutes, which is itself revealing — most people have never seen their own memory dossier laid out this way.
Edit happens in Memory.md. Delete the stale stuff. Keep the parts that genuinely describe your preferences, working style, and project context. Compress. Reorganize. The file is yours, and it stays a file.
Push sends that cleaned-up memory into the next assistant you want to use. What changes is the model. Not your context.
I want to be honest about one thing: the current version is a human-in-the-loop system. The extension opens the target platform, injects the sync prompt, and tracks what's been synced where — but you review and confirm each step. I could have built something flashier and fully automated. I chose not to, because opacity is exactly what I was trying to solve. You should see what's being moved. At minimum until the trust is built.
What I've noticed since I started using my own tool
A few things have changed for me, and they might change for you.
My Memory.md is shorter than I expected. Once I actually looked at what the platforms had saved, maybe thirty percent of it was worth keeping. The rest was noise — one-off comments, stale project context, assumptions inherited from a mood I was in six months ago. Seeing it laid out made it trivial to cut.
My outputs got a little more surprising again. Not better every time. But I'm no longer always being told what I already want to hear. I trust the pushback more when it comes, because the model has less of a profile to flatter.
And I stopped feeling locked in. When I decide to use Claude for a long writing session instead of ChatGPT, it no longer feels like I'm abandoning a relationship. I'm just switching tools. The context travels with me.
The point isn't to replace platform memory
I want to be clear about what Memory Sync is and isn't.
It isn't an attempt to replace what OpenAI, Anthropic, and Google are building internally. Their memory systems will keep improving. Some of them — Claude's project-scoped memory, for instance — are already meaningfully better than the global-dossier approach. That trend will probably continue, and I'm happy about it.
What Memory Sync is, is infrastructure for the part these companies have no incentive to build: the layer between them. The Memory.md that sits in your hands and travels across platforms. The audit trail of what's being remembered about you. The portability that turns your long-term context into something you own instead of something you rent.
I think you'll be surprised how much of it shouldn't have been there in the first place.
]]><![CDATA[Some high-DR personal branding websites (Free)]]>https://www.zhengyang.me//high-dr-pd-website2025-02-19T16:00:00.000Z
This is some free high-DR personal branding websites to help you build your personal brand and provide high-quality backlinks for your projects. Enjoy~
]]><![CDATA[A Simple Guide to SEO]]>https://www.zhengyang.me//a-simple-guide-to-seo2025-02-12T16:00:00.000Z
SEO is crucial for improving your website's visibility in search engines. Here's a comprehensive guide covering the key aspects of SEO.
Keyword Research
Large Language Models (LLMs) are currently the most effective tools for keyword research. They can help identify high-volume, moderately competitive keywords for your content.
💡
PROMPT: Please provide some high-search-volume, moderately competitive short-tail and long-tail keywords based on the information from the website above.
On-Page Optimization
Here are the essential elements of on-page SEO:
Title Tags: Include primary keywords in page titles while maintaining click-appeal
Meta Descriptions: Craft concise, engaging descriptions that incorporate relevant keywords
Keyword Strategy: Focus on both:
Short-tail keywords (e.g., "SEO")
Long-tail keywords (e.g., "how to implement effective SEO")
Long-tail keywords typically face less competition and offer better ranking opportunities
Content Optimization: Naturally integrate keywords into your content while avoiding keyword stuffing
Header Tags: Structure content using H1-H3 tags properly. Use only one H1 per page, with H2 and H3 for subsections
Image Optimization: Add relevant ALT text and use keyword-optimized file names
Internal Linking: Create a network of internal links to improve crawl depth and user experience
External Linking: Link to authoritative websites to build credibility
Schema Markup: Implement structured data like Product and FAQ schemas to enhance search visibility
Technical SEO
Key technical optimization aspects include:
XML Sitemap: Create and submit sitemaps to help search engines efficiently discover your important pages
Robots.txt: Configure properly to control crawler access and optimize crawl budget
Structured Data: Implement Schema.org markup to help search engines better understand your content
By following these guidelines, you can create a solid foundation for your website's SEO strategy and improve your search engine rankings.
Export and Manage AI Conversation History – Optimize your search and review experience with support for popular AI models (such as Deepseek, ChatGPT, Claude, Gemini, Perplexity, Poe, and more).
The "Coming Soon" webpage went live in four hours on Jan 15, 2025.
Requirement research and product incubation in Februray.
Reasons for failure: Initially, this project was set up just to chase trends, but after extensive research on Reddit, no worthwhile demand was identified. Although some tools could be developed, such as a tool to help foreigners choose Chinese names or a web-based Xiaohongshu translation tool, these needs were merely minor itches, not pain points, and couldn’t provide paid value. In the future, some of these user itches might be turned into hackathon projects, but the goal of "12 startups in 12 months" is to create profitable projects. Additionally, last month’s project took up too much time, so it ended in failure.
失败原因：起初这个项目只是为了追赶热点而设立的，但是在reddit上调研许久都没有发现值得去做的需求点，尽管可以做一些工具例如：帮助外国人起中文名字的工具、网页版小红书翻译工具。但这些需求只是一些痒点，并不是痛点，提供不了付费价值。后续可能会把一些用户痒点作为黑客松项目来实现，但是12 startups in 12 months希望的是做一些可盈利的项目再加上一月的项目占用了太多的时间，因此这个月的项目以失败告终。
The MCP Server designed to guide Vibe Coder's development.
CodeZap is an MCP Server, helping Vibe Coders take the crucial step from 0 to 1 in product development.
MCP Server
First commit on Apr 18, 2025.
Landing Page
First commit on Apr 22, 2025.
❤️Some feelings
Recording some of my thoughts and feelings during the development process.
Having many ideas is easy, but turning an idea into a truly great product is not. (Mar 3, 2025)
As an entrepreneur, one should lean more toward sales rather than technology, and especially avoid becoming obsessed with technology. Since I only started dabbling in UI/UX after embarking on entrepreneurship, my "OCD" self, who considers my aesthetic sense to be excellent, often gets hung up on certain components of the product and website, resulting in extremely slow progress. This is highly detrimental to entrepreneurship. Instead, energy should be focused on product functionality and promotion—f**k the UI, as long as it’s usable. First, get the product functionality right; once you have users, then refine the UI/UX. This will improve user retention and increase organic sharing rates. (Mar 26, 2025)
