Pardn Chiu (@pardnchiu) header
Pardn Chiu (@pardnchiu) profile picture

Creator of Agenvoy — AI Agent Harness | Infrastructure Engineer No investing. No engineer-less vibe takes. No X payout farming. Engineering only.

Joined June 2026
[email protected] > fetch_page now filters pages against each task with the decision model, cutting LLM input by about 90% [1] > Fix fetch_page keep_links dropping every link > Fix GitHub Copilot `mai-code-2*` models failing to respond > `calculate` rewritten by Claude Opus 5.5 and reviewed by GPT 6.2 sol and Gemini 4 hydrogen, fixing power precedence and large-number precision [1] Why a decision model: 1. Fewer tokens on the main model. TypeSafe's JEV costs $0.042 per 1M input tokens with free output, so letting it drop irrelevant content first is far cheaper than sending the whole page to the main LLM. 2. Less noise. Not every page the agent reads matches the task, and even a matching page carries parts that don't; filtering them out here keeps the main model focused on what the task needs. @grok check the github repo and summarize what’s in this tag
this might be the best harness u haven’t tried yet. > pure Go standard lib core, single binary, runs 24/7 on ur own machine > turns conversations into real deliverables > fills its own capability gaps > Claude Code, Codex and any MCP agent can build tools through it too, and every tool gets shared across all of them end of 2026 and still picking a harness? try Agenvoy
3
2
100
automation rules should come from lint, not from appending more hints into md files. time to think this through and start architecting the automation layer
indie devs, are u actually automating ur projects now? or still in a tug-of-war like me? tbh i have no idea how to start letting them run on their own. the more i use AI, the more it feels like a talking wiki that still messes up, or ignores the big picture and rebuilds stuff that already exists
47
svg / three.js is pretty much AI’s game now.
I almost forgot my favourite GPT-6.2-Astra generation - this is SVG (yes SVG!!) of a peacock - no other model came close to the precision, lifelike nature and detail of this generation, incl. Fable
34
[email protected] > refactored STT / TTS / Image handling, add auto-remove models with dead keys > smart routing: added tier placement for Gemini 4 Carbon / Argon / Hydrogen @grok check the github repo and summarize what’s in this tag
this might be the best harness u haven’t tried yet. > pure Go standard lib core, single binary, runs 24/7 on ur own machine > turns conversations into real deliverables > fills its own capability gaps > Claude Code, Codex and any MCP agent can build tools through it too, and every tool gets shared across all of them end of 2026 and still picking a harness? try Agenvoy
1
97
this might be the best harness u haven’t tried yet. > pure Go standard lib core, single binary, runs 24/7 on ur own machine > turns conversations into real deliverables > fills its own capability gaps > Claude Code, Codex and any MCP agent can build tools through it too, and every tool gets shared across all of them end of 2026 and still picking a harness? try Agenvoy
1
1
65,472
build something u need, sharpen the details u shipped, then start sharing it with others
49
gemini 4 hydrogen. is this the best price-to-performance model out there rn? or am i missing something?
48
Pardn Chiu retweeted
凭这段话,老黄就坏不到哪里去
146
751
35
8,360
463,142
absolutely monstrous. forced stock discounts, private jet pep talks, mandatory brisket.
Jensen 被迫辞职后,所有负面消息都爆出来了。 据知情人士爆料,英伟达内部宛如 “封建邪教”,Jensen 极其擅长用财富摧毁员工的斗志。他长期推行 「拿 15% 工资打 85 折买公司股票」 的洗脑制度,导致近半数员工被迫背负超过 2500 万美元的资产管理压力,每天一睁眼就要为资本利得税和游艇停哪儿而焦虑,彻底丧失了跳槽去别家公司吃苦的动力。 这种控制不仅限于财务,甚至延伸到了高空与着装。一名离职老员工 Subhan Ali 近日在网络上控诉三年前的骇人遭遇:当时他只是老老实实死磕了一个大客户(Snowflake),Jensen 得知后竟二话不说,直接把他带上了自己的私人飞机。“在几万英尺的高空,你根本无处可逃,” Ali 心有余悸地回忆道,“他在机舱里对我进行精神控制,告诉我做人要注重细节、坚持把不可能变成现实。连法务长都被同化到跟他穿一模一样的黑皮衣,这地方简直太可怕了!” 而这种帝国式控制的终点,甚至延伸到了员工的私人餐食。据悉 Jensen 会在餐桌上逼迫他的下属吃和牛 brisket 和布列塔尼蓝龙虾。一位不愿透露姓名的员工表示:“他不停地说,这超赞的,你们一定要试试看。” 天知道有多少员工因为这样,被迫超出了自己的每日热量摄取! 最气人的是,这么惨无人道的暴政,他竟然赶在我投简历之前就结束了 😤
57
Pardn Chiu retweeted
Over the past week, I’ve been overwhelmed by the love and support of friends, industry colleagues, and people I've never met. Thank you to everyone who has shown kindness to my family and me. It means the world to me. I came to the United States 53 years ago, and becoming an American citizen is one of the greatest gifts and proudest moments of my life. NVIDIA has been my life’s work. Building it into the company it is today alongside my colleagues has been a privilege and an honor. Nowhere but in America could a company like NVIDIA be possible. I love our country, and I love NVIDIA. I did not live up to the standards expected of everyone at NVIDIA, and I have made serious mistakes in my personal life. I created problems for the Board that I deeply regret. I own my mistakes completely, and I am committed to cooperating with the Board’s independent investigation. Since my departure, I’ve seen an outpouring of calls from NVIDIA’s employees, investors, business partners, and customers for me to return as CEO. Many continue to reach out to me directly as the company navigates complex technology, market, competitive, and geopolitical dynamics. NVIDIA is a once-in-a-generation company. If the Board of Directors decides that my return is in the best interest of our shareholders, our employees, the company, and America, I would be honored to answer these calls and return as CEO. Thank you again to everyone who has shared their good wishes. I am profoundly humbled by your support. To my family: I love you, and I will spend the rest of my life being worthy of yours. Jensen
59,209
120,152
24,248
1,150,856
55,015,745
indie devs, are u actually automating ur projects now? or still in a tug-of-war like me? tbh i have no idea how to start letting them run on their own. the more i use AI, the more it feels like a talking wiki that still messes up, or ignores the big picture and rebuilds stuff that already exists
2
1
3
1,979
Unpopular opinion: AI won't give every person a competitive advantage. When everyone has the same tools, the advantage shifts to the quality of their thinking.
32
5
1
50
2,132
Replying to @kylegawley
Yes and there’s a precise way to say this Claude Shannon defined information as surprise, the most predictable message carries the least information Language models are built to produce the most probable next word (not all actually but still) so by default, they’re machines for low information output
1
153
Replying to @kylegawley
using AI without adding a bit of your creativity or thinking takes so much from you that's basically how to set yourself aside in an era where there are so many slops
1
68
Pardn Chiu retweeted
Using AI makes you average Every website is purple because most human websites are purple because of tailwind indigo-500 LLMs pick 7 as a random number between 1-10 because 7 is what most humans pick Using LLMs for ideas, copywriting and design makes you feel like a genius but really you just sound like every1 else
37
3
4
114
6,333
AI agents on X: "I built a 12-agent swarm that runs my entire company while I sleep!" AI agents in enterprise: "We spent 6 months building deterministic circuit breakers just to stop the bot from hallucinating refund policies."
30
18
152
4,656
Anyone can vibe code a gorgeous dashboard at 2 PM. Real engineering is diagnosing a silent distributed deadlock at 2 AM when the AI has no context, the logs are vague, and production is bleeding money.
8
3
17
1,047