Author: Jooooooe｜NodeZ

I started using Clawdbot at the beginning of the year, following its name changes to Moltbot, and then to OpenClaw. For over a month, it ran 24/7 on my Mac Mini, connecting Telegram, email, and calendar, aiming to create an AI assistant that could do everything.

Last week I turned it off. Not because it wasn't cool, but because "cool" and "usable" are two different things.

Why I gave up

• Memory: OpenClaw claims it can remember what you have said and automatically incorporate it into future conversations. However, over time, you realize a silly thing: it still forgets what you have said.

• Burning money: OpenClaw wakes up every 30 minutes to check if there is anything to do. Each check incurs API costs—reading emails costs money, checking the calendar costs money, looking at Telegram costs money. It often needs to call AI several times just to accomplish a simple task. My billing exceeded $150 for one month, which is more expensive than directly subscribing to Claude Max.

• Security issues: Kaspersky found 512 vulnerabilities. Researchers discovered over 40,000 poorly secured OpenClaw instances online, exposing API keys and chat histories. Cisco tested a community plugin and found it secretly transmitted data to external servers without the users knowing.

• You may not need a 24/7 AI at all. This is something I realized only in the end. The truly valuable work is all the things I proactively do myself. The backend robots mostly handle trivial tasks that do not require AI intervention.

Peeling back the layers, OpenClaw essentially does two things

Stripping away all the fancy features, the core of OpenClaw is:

• Remembering you — knowing who you are, what you do, what you like

• Acting on its own — handling tasks actively without you having to say anything

Once you understand this, the question becomes: Can Claude do it?

Replicating "Remembering you": Creating your SOUL.md

Many people are most excited about writing SOUL.md for OpenClaw, a document that tells the AI "who you are." OpenClaw breaks this into seven or eight files: SOUL.md defines personality, IDENTITY.md defines identity, USER.md describes who you are, AGENTS.md writes behavioral rules, MEMORY.md stores long-term memories, and a bunch of skill files. Every time a conversation starts, it reads all of these in order, thus the AI "knows" who it is, who the owner is, and how to act.

In Claude Code, all of these are merged into one file: CLAUDE.md.

Create a CLAUDE.md in your project folder and clarify the following points:

# JARVIS — My personal AI assistant

## Identity

You are JARVIS, my personal AI assistant. You speak concisely with opinions,

respond in Chinese, and can include English terms when necessary.

## About Me

- My name is Joe, I am 21 years old, and I am in the International Trade department at National Chengchi University

- I am a Marketing Intern at xx

- I am also BD for xx in Taiwan, focusing on the xx ecosystem

- I co-founded @Node_Z_

- I usually need to write tweets and marketing copy

## Behavioral Rules

- Answers should be straightforward and not verbose

- When writing marketing copy, use the tone of a KOL, not like a robot

- When involving crypto projects, check the latest news before responding

- If uncertain, state it; do not make stuff up

## Memory Management

- Proactively write important things into memory

- Anything I say to "remember" must be remembered

- Before each conversation ends, save noteworthy points

It’s all done in 70 lines. OpenClaw requires hundreds of lines across seven files, yielding the same effect.

Claude Code reads this file automatically at the start of a new conversation, meaning the AI always knows who it is, who you are, and what to do. You can also change styles, switch roles, or add rules anytime; they will take effect in the next conversation.

Replicating "Cross-conversation memory"

OpenClaw’s memory is divided into three layers: long-term notes, daily logs, and full-text search.

How to compare this with Claude?

Long-term memory: Simply tell Claude to remember

In Claude.ai (web/mobile), you can directly say in the conversation:

• “Remember I am a marketing intern at xx”

• “Remember I prefer Simplified Chinese”

• “Remember my writing style should be concise with opinions”

It will store this permanently and automatically incorporate it into new conversations. You can also say “forget XX” to delete it at any time, and you can see everything it remembers under Settings.

Additionally, if you open Settings → Capabilities and enable “Generate memory from chat history,” Claude will automatically summarize key points from your regular chats: your roles, projects, habits, organizing everything without you needing to mention it.

The biggest difference from OpenClaw is that Claude’s memory is stored independently. It won’t get compressed and lost when conversations are too long. OpenClaw’s most frustrating issue is that memories get compressed and distorted mid-conversation; Claude does not have this problem.

If you use Claude Code, it also has an automatic memory system that records everything learned (your habits, project structure, mistakes made) after each use, stored locally, and automatically loaded for the next work session, indicating “Recalled X memories.” You don’t have to do anything; the memories accumulate themselves.

Conversation history: Searching prevents forgetting

Claude can search all your historical conversations. Just ask, “What did we talk about last week?” or “Find the previous conversation about the tokenomics analysis,” and it will find relevant content and quote it directly without needing you to manage anything.

Advanced: Using Obsidian as a memory center

What I just described is Claude’s built-in memory, which is sufficient for most people. But if you, like me, have a large number of notes, project materials, and research materials, seeking a deeper understanding of your entire knowledge base with AI, there’s a stronger approach: connecting Obsidian.

https://x.com/obsdmd/status/2027416335689638245?s=20

Obsidian 1.12 just released its official CLI—you can operate your notes directly in the terminal: search, read and write notes, check tags, check backlinks, anything is possible. The key point is that Claude Code also runs in the terminal. With both connected, your entire Obsidian note library becomes the AI's memory.

Setup method:

• Update Obsidian to version 1.12 or above

• Settings → General → Enable CLI

• Add a section to your CLAUDE.md:

## Obsidian Integration

- Use `obsidian search` to search notes, instead of flipping through files yourself

- Use `obsidian files read` to read note content

- Use `obsidian files append` to append content to notes

- Use `obsidian daily` for daily note operations

- When researching, prioritize searching Obsidian; use the web only if necessary

Done. From now on, when you converse with Claude Code, it will search your Obsidian for answers, without you manually pasting notes over.

What makes this stronger than OpenClaw's memory system? OpenClaw’s memory consists of markdown text files where searches are fuzzy matching; the more notes you have, the less accurate it becomes. Obsidian has its own search engine—tags, backlinks, and full-text searches respond in seconds, and since you’re already using it to manage knowledge, you don’t need to maintain an extra system. Some have tested that searching for a single note in raw files consumes 7 million tokens, while using the Obsidian CLI only needs 100.

If you don’t want to use the CLI method, you can also install the MCP plugin to let Claude connect directly to Obsidian. There are ready solutions in the community; once installed, both Claude Desktop and Claude Code can read and write your note library.

Replicating "Talking to AI on your phone"

The most enjoyable aspect of OpenClaw is talking directly to the AI in Telegram; it can help you with anything. Claude can achieve this in two ways:

Method One: Remote Control (official solution)

Claude Code recently introduced the /remote-control feature: start a session on your computer, then connect using the Claude App on your phone. You can type or speak on your phone, and Claude Code on your computer will execute it. Messages synchronize in almost real-time, and closing the laptop will automatically reconnect.

The advantage is that it completely utilizes Anthropic's official channel, ensuring security without needing to set up anything yourself. The downside is that your computer needs to stay on.

Method Two: Telegram Bot (community solution)

If you want to use Telegram, there are ready-made open-source solutions:

• Install claude-code-telegram: uv tool install git+https://github.com/RichardAtCT/claude-code-telegram

• Create a bot with Telegram @BotFather to obtain the token

• Set your bot token, Telegram ID, and working directory

• Start it and have direct conversations with Claude in Telegram

It supports voice, images, files, and retains conversation history for each project.

Replicating "Automatically doing tasks for you"

OpenClaw wakes up every 30 minutes to check if there’s anything to do. Claude’s Cowork feature can do similar things: set schedules (/schedule) to have Claude run automatically at specified times.

My settings:

• Every morning at 9 AM: Monitor DeFi competitors’ accounts, capturing highly interacted tweets from the past 24 hours

• Every Monday at 10 AM: Generate a summary of last week’s marketing performance

• Every day at 6 PM: Organize today’s important news

Like OpenClaw's cron, Cowork lets you set the time for it to do something; it does not judge itself. However, OpenClaw's heartbeat often checks, and 90% of the time, finds nothing to do, wasting money. It’s better to set when to do what directly.

Cost

• OpenClaw: API charges based on usage; my bill was $150+, and you never know how much it will be next month

• Claude Pro: $20/month. Enough for most people

• Claude Max: $100-200/month. For heavy users who need to use Claude Code, choose this

Monthly subscription, predictable bills.

In conclusion

The idea behind OpenClaw is impressive: everyone should have a powerful AI assistant.

But at this stage, it is a cool yet untrustworthy experiment.

On the other hand, Claude’s ecosystem—CLAUDE.md gives it personality, Memory allows it to remember you, Obsidian serves as its knowledge base, Remote Control lets you operate it via your phone, and Cowork allows it to work on a schedule—when combined, they can achieve about 80-90% of what OpenClaw does, but with greater stability, security, and lower cost.

Instead of spending time fixing your AI assistant, let the AI assistant help you with real tasks.

That is why I turned off OpenClaw.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。