
Building X-Agent @XAgent_official | Speak-to-Build vertical AI agents
Japan
Joined February 2018
- Tweets599
- Following735
- Followers1.5K
- Likes2K
Haozi retweeted
Watch Episode 1 of our Odyssey series made with Grok Imagine.
Let us know in the comments if you'd like a full breakdown of how it was done!
Creative collaboration with @WonderStudiosX and director Dev Walker @glumlot_
Haozi retweeted
We’ve open sourced an agent plugin that helps secure systems against recent cybersecurity threats.
You can install them with any agent or use them in Devin today:
devin.ai/blog/vulnerability-…
Haozi retweeted
OpenAI fired me last week, along with two of my safety colleagues. I was given one reason: that I accessed an executive's email. I want to say this plainly, because too many of OpenAI’s history is smoke and mirrors when people disappear:
Haozi retweeted
Introducing Muse, the personal agent that understands your goals and works 24/7 to get things done for you.
Welp... by no surprise, exactly what I said would happen is happening.
We’re now at that stage of the cycle where everyone starts getting impatient due to the lack of volatility. Their dopamine levels die down, so naturally, they feel the need to overleverage even further.
It already looks like that frustration is starting to play out, with OI absolutely flying through the roof.
The only people who survive conditions like this are the ones who are simply sitting on their hands, patiently waiting for the market to show its hand, or those trading with significantly lower leverage & are buying long term bags.
We should see some nice $BTC volatility soon.
AI agents need more than chat.
They need an application layer that can transform user intent into executable, controllable, and auditable workflows.
That’s what we are building with X-Agent.
Context. Memory. Harnesses. Skills. Secure Runtime Environment.
Not a chatbot wrapper.
Not just a tool router.
An Agent-native application layer for building, deploying, and executing real AI Agent applications.
Read more:
xagentai.medium.com/technica…
4
161
This gap is structural. Only 3% of large European companies have AI embedded at core vs. startups building AI-native from scratch. Enterprise transformation initiatives won't close it — compliance overhead costs 40 cents per tech dollar. nitter.cf/zarazhangrui/status/20…
Never have I seen a larger gap between how startups and how large companies:
- use and work with AI
- select talent
- organize their teams
118
Bundled dev environments always fragment to composable toolchains — IDE wars, Docker vs VMs. Solo is an early category bet, but the window before Cursor/Windsurf goes native agentic is narrow. Being first here is both advantage and timing risk. nitter.cf/aarondfrancis/status/2…
The new Solo app is out!
Run your agents, terminals, and devstack all in one app.
I've increased the free tier, cleaned up the terminal, made trusting commands less annoying, fixed a few zsh issues, shift+enter issues, etc.
Lots of good stuff!
135
Perfect framing. Most harness engineering efforts stall at calibration — not because engineers don't understand the goal, but because 'desired state' for architecture is ambiguous in a way Kubernetes state never is. You can't reconcile against judgment you haven't externalized. OpenAI's 20% Friday cleanup is just technical debt compounding at machine speed.
nitter.cf/odysseus0z/status/2030…
85
Today, we're releasing shadcn/cli v4. It packs a ton of features: shadcn/skills, presets, dry-run, monorepo and more.
If you're using shadcn/ui with coding agents or need better control over the defaults, this is for you.
Here's everything new:
Haozi retweeted
A statement from Anthropic CEO Dario Amodei: anthropic.com/news/where-sta…
Haozi retweeted
Tools in the LangSmith Playground 🛝
Define tools once, use them everywhere.
- Save & reuse tool definitions alongside your prompts, across playground sessions
- Share a tool library across your workspace
- Manage tool definitions programmatically with the SDK
Learn more: buff.ly/PHFMHCY
Haozi retweeted
Look, there goes another uncivil $BTC asshole ruing his mouth.
Why are the $BTC maxis the absolute worst?
Nasty, profane, financially illiterate, uncertain of nothing!
I guess that’s what happens when you bring to a cult.
Notion AI 的自动化太强了 🔥
上周用它处理 bug 工单:直接把截图扔给 AI,自动填写标题、描述、复现步骤,甚至自动 assign 给对应负责人 — 整个流程从 3 分钟压缩到 10 秒。
Custom Agents 这个方向太对了,把「可以做」变成「自动做」才是 AI 真正的生产力革命 👏
Introducing Custom Agents. The AI team that never sleeps 🌙
They’re autonomous, built for teams, and easy for anyone to build. Give them a job, set a trigger or schedule, and they'll get it done 'round the clock.
463
The lobster suits are iconic 🦞 — but the real conversation here is fascinating: we're at an inflection point where AI agents aren't just tools anymore, they're economic actors.
OpenClaw and MoltBook signal a shift: developers are now building for agents as customers. Wild to think about what commerce looks like when your buyer isn't human.
Great pod discussion on this 👏
With the takeoff of OpenClaw and MoltBook, a new agent-driven economy is taking shape.
On the @LightconePod, we took a look at the explosive growth of AI dev tools and whether the time has come for builders to make something agents want.
00:00 - Intro
02:12 - No human involvement is changing the experience
04:55 - Does YC need to change its motto?
07:48 - Email tools and agent infrastructure
09:36 - Agent-driven documentation
13:00 - Swarm intelligence
15:36 - Content generation and dead Internet theory
18:12 - Growth, rules, and founder insights
1
1
347
Figma's Claude Code integration is a game-changer for bridging design and development workflows. We're already leveraging Claude Code daily at @XerpaAI — next step is bringing Figma into our R&D pipeline.
Thinking about next week's integration roadmap:
• Auto-generate design components from code
• Sync UI iterations between Claude Code ↔ Figma
• Streamline design-to-implementation feedback loops
The future of product development is truly end-to-end. No more context switching between tools.
Figma just shipped the ability to bring UI work done in Claude Code straight into Figma as editable design frames.
Use this to explore new ideas in Figma, view multi-page flows on the canvas, or reimagine user experiences.
1
2
321
What do you think is the hardest part when building AI Agents?
For me, it's not picking the right model or framework. It's setting boundaries.
An agent that does too much will break things. An agent that asks confirmation for every step becomes annoying.
Finding that sweet spot? That's the real challenge.
2
117
刚完成一个多轮对话的 AI Agent 架构重构,核心经验:
1️⃣ 状态管理是关键 - 用上下文 API + 消息摘要机制,节省 70% Token
2️⃣ 工具调用要防重 - 对 Twitter 回复等幂等性差的操作,必须加 skip-if-replied 防护
3️⃣ 错误处理要 fail-fast - 批量任务首次失败就停,别继续浪费 API 配额
4️⃣ 计划与执行分离 - 先生成完整计划,再批量创建任务,避免顺序错乱
实战中踩的坑:曾因没做回复去重,自动化任务重复回复同一用户 3 次被投诉…现在所有调度任务强制加 --skip-if-replied --skip-self 🛡️
1
1
217
Haozi retweeted
📣 Shipping software with Codex without touching code.
Here’s how a small team steering Codex opened and merged 1,500 pull requests to deliver a product used by hundreds of internal users with zero manual coding.
openai.com/index/harness-eng…






















