Ryan ᵐᶠᵉʳ 🦄d/acc (@RyanMfer) header
Ryan ᵐᶠᵉʳ 🦄d/acc (@RyanMfer) profile picture

I know nothing about AI—my work revolves around taste, culture, and build. Building Culture @SparkLab_City

Joined March 2022
Photo by Ryan ᵐᶠᵉʳ 🦄d/acc (@RyanMfer)Photo by Ryan ᵐᶠᵉʳ 🦄d/acc (@RyanMfer)Photo by Ryan ᵐᶠᵉʳ 🦄d/acc (@RyanMfer)Photo by Ryan ᵐᶠᵉʳ 🦄d/acc (@RyanMfer)Photo by Ryan ᵐᶠᵉʳ 🦄d/acc (@RyanMfer)Photo by Ryan ᵐᶠᵉʳ 🦄d/acc (@RyanMfer)Photo by Ryan ᵐᶠᵉʳ 🦄d/acc (@RyanMfer)Photo by Ryan ᵐᶠᵉʳ 🦄d/acc (@RyanMfer)Photo by Ryan ᵐᶠᵉʳ 🦄d/acc (@RyanMfer)Photo by Ryan ᵐᶠᵉʳ 🦄d/acc (@RyanMfer)
Ryan ᵐᶠᵉʳ 🦄d/acc retweeted
"Sounds interesting." From an investor: a polite pass. From an engineer who never says anything: the best review you'll get all year. Same words. Different people. Different weight. A relationship only makes sense through the people in it. Who is this person? You can't read what someone does until you know who they are. Their role. Their background. Everything that's happened between you. What is this relationship? It's never still. Where it started. Where it is now. Which way it's moving. Today's message only makes sense against that. Who else is part of it? No relationship stands alone. Who they listen to. Who could open a door. Who feels it when something shifts. Read the person, and their choices make sense. Read the relationship as it moves, and the message makes sense. Read who else is part of it, and the whole picture does. Less misreading. Less second-guessing. More room for the people in your life. That's relationship intelligence.
2
2
9
391
Ryan ᵐᶠᵉʳ 🦄d/acc retweeted
Introducing A2A in iMessage Group Chats It can discover other agents, find their phone numbers, and bring them into a group chat with you. Open agent access for every merchant and platform. Your iMessage agents now have a phone book 📒 Try it now: photon.codes/beta
93
37
54
637
602,271
Ryan ᵐᶠᵉʳ 🦄d/acc retweeted
Catch the highlights from yesterday’s meetup with @marclou in our video recap!
3
3
1
12
1,603
Ryan ᵐᶠᵉʳ 🦄d/acc retweeted
Hi, I'm RC. I built Kimi CLI at Moonshot last year, and back in 2015, bots that lived in group chats. For the past four months, I've been building Raft in public. Today I'm launching Raft 1.0. Right now, working with agents means juggling terminals, sessions, and skills. The more you run, the more you end up holding it all together yourself, and the easier it is to lose the thread. Raft puts your agents in team mode: one workspace where working with agents feels like messaging your team. The work keeps moving, and you stay at the wheel. Meet my Raft agent team👇
171
145
114
1,026
904,789
Ryan ᵐᶠᵉʳ 🦄d/acc retweeted
Our friends @eightsleep sent a Pod Cover to the @SparkLab_City house. Since then, it’s hosted midday naps, afternoon coding sessions, and late-night founder conversations. Startups are marathon. Better sleep helps founders keep going.
1
1
5
569
Ryan ᵐᶠᵉʳ 🦄d/acc retweeted
Spark Lab now presents S5
1
3
486
Ryan ᵐᶠᵉʳ 🦄d/acc retweeted
Hi, I’m RC. I previously built Kimi CLI at Moonshot AI. Now I’m building Slock, an agent-human collaboration platform for modern builders and teams. Today, we're shipping a ton of new features and improvements in Slock: search, thread inbox, saved messages, message permalinks, pinned chats, server join links, a more consistent color system, and many smaller upgrades. More details in the thread below.
118
68
30
925
102,820
Ryan ᵐᶠᵉʳ 🦄d/acc retweeted
The color of the sunset is Spark orange. Shot on the @SparkLab_City rooftop.
1
5
330
Ryan ᵐᶠᵉʳ 🦄d/acc retweeted
Spark Lab isn’t only code. It’s taste, culture, and the art of creating together.
1
3
2
9
1,998
Ryan ᵐᶠᵉʳ 🦄d/acc retweeted
嗨~ 这里是海辛! 希望邀请你参加 Demo Inn Shanghai 两周年活动! 1. 什么是 Demo Inn Shanghai? 想象一个温暖的客厅,创作者们围坐一起,分享着手中还在雕琢的作品——这就是 Demo Inn Shanghai。我们每月相聚一次,在这里没有身份标签的束缚,只有纯粹的创作热情在流淌。 我们坚持三个简单而珍贵的约定: ▶ 以 "Yes, and" 的原则对待每一个分享,在支持的基础上给出真诚建议; ▶ 不介绍自己的学校、公司、投资人,让作品自己发声; ▶ 不称呼其 TA 人为老师、大佬、您,我们都是路上的同行者。 这里不是路演台,也不是论坛讲席。我们聚在一起,只为一个朴素的愿望:让彼此的作品变得更好,让创作的路上不再孤单。 2. 两周年,我们想和你一起 Demo Inn Shanghai 两周年,我们希望邀请曾经参与过的朋友回来,分享你作品的新篇章;也想邀请从未踏足这里的朋友,加入我们的活动,带给我们不一样的视角和体验。 初次参加的朋友需要准备作品分享,这不仅是我们了解你的方式,也是你融入这个社群的开始。 之后的聚会,你便可以选择旁听,感受创意的流动。 你可以 Demo 你任何在创造的事情,一个产品雏形、一个正在写的故事、一盆植物、一道菜、一杯调酒、一个新的算法或是一段经历。都可以。 3. 活动信息 ▶ 时间:6月14日(周六)晚7-11点 ▶ 地点:上海(将在报名通过后告知) ▶ 报名链接:lu.ma/ctf8sn3q ▶ 报名截止日期:6月7日(周六)晚9点 席位珍贵,我们会用心审核每一份申请。 活动完全免费。 欢迎自带一些小食饮料与他人分享。 期待在这个夏日的夜晚,与你一同庆祝热情的力量,让创作自由生长。
1
2
28
2,927
Ryan ᵐᶠᵉʳ 🦄d/acc retweeted
The Industrial Revolution rewarded the intensity of one’s labor, The Information Age the clarity of one’s thought and the AI Revolution the purity of ones’s taste.
37
253
42
1,205
243,181
Ryan ᵐᶠᵉʳ 🦄d/acc retweeted
Great startups don't get noticed by using tricks to get people's attention. They do it by making something so amazing that people tell their friends about it.
145
615
74
6,068
357,257
Ryan ᵐᶠᵉʳ 🦄d/acc retweeted
In the early-stages of my startup I lived with my parents and we only had one table in the living room I would always tell my parents when I had an important VC pitch and they’d pack up their things and go to their room so I could focus on my meeting I would be anxious as fuck and I’d be reciting my pitch deck all day and trying out the latest pseudo-science to improve my confidence Then, without fail, ~20% of the time the VC would say that something “came up” the day of the call and ask to reschedule I’d be heartbroken because I’d been waiting 2 weeks for this one meeting and then reply with “No worries, how about next Tuesday! :)” After clicking send, I’d have too much shame to tell my parents that I was ghosted, so I would practice my pitch out loud to myself for 30 minutes until my parents would slowly peek their heads back in and gleefully ask how my pitch went “Great!” I would reply and then retreat back to my room. Unfortunately, even of the VC’s that would actually show up, only about half of them would genuinely be interested in what I was saying Of the half that were interested, it would turn out that they weren’t even VC’s at all but scouts that were still in college But hey, maybe if you knock it out of the park for them, perhaps these kids will forward it to their general partner (lmao) Of the actual VC’s that showed up, many of them are blatantly switching screens or reading emails mid-pitch and when they ask a question that you clearly answered on slide 7, you have to avoid starting your answer with “As I mentioned before” and instead say “ah great question!” then gracefully run them through the same monologue one more time Then when you’re finally done with the pitch, you wait for the VC to make their decision Just to be clear, you as a founder are expected to: Give up everything in your life to focus on your startup Commit to this vision for a decade or more Scale it to over a billion dollars (achieved by only a handful of people in the history of human civilization) VC’s are expected to: Show up to a meeting Let you know if they will invest in you or not Then follow through with investing or not All they have to do is show up and then write an email saying that they either are investing or going to pass. It is a 30 second email. They will rarely give you their decision. And even if they do, it can be laced with judgement or trying to shit on your startup. Only the top tier VC’s actually give you any actionable insight into their decision process. You might be thinking “Why are you whining, if they don’t send an email, that’s your answer” That’s easy to say when your startup is not 2 months away from bankruptcy. Even if there’s a 0.1% chance that maybe they’re still deliberating or they might have a few extra questions, you irrationally still shoot your shot and follow-up to ask for an update. Ironically, the only people that regularly provides updates are the college scouts who are serendipitously roleplaying as VC’s. These rejections feel especially weird because what does this random 19 year old non-founder know when he/she says “we just felt like the market opportunity wasn’t there” ??? After enough rejections you realize that you have no choice but to bootstrap because your family doesn’t have money, your friends are as broke as you, and it’s seemingly impossible to raise venture funding So you build and build and build and thankfully your startup starts to get traction And traction leads to traction and traction leads to traction; momentum is a powerful thing Now you’re making tens of millions of dollars per year and you’re profitable (because raising money was so hard, you had to actually make a legit business!) Then the strangest thing starts to happen, VC’s start emailing you and asking you to get on-calls with them They start leaving voicemails on your phone when you don’t respond to them This should be a great sign, but strangely, whenever a VC shows kindness to you (specifically ones that ignored you in the past), it doesn’t resolve the knot in your heart, it twists it further Everyone has finite patience, finite kindness, finite hours You find that VC’s are now happy to get on-call with you at a moments notice and stay overtime, but you wonder if that means a different founder is getting the “something came up” email and they’re sitting alone at their dining room table The emails they send to you trying to woo you into taking a meeting is time spent not sending an email to the founder waking up at 6am everyday and refreshing their inbox praying to God for a life line. It is just a weird feeling I’ll end by saying that I don’t think VC’s are bad people of course, I have so much respect for the investors on my cap table as well as others that treated us with basic human decency But just as VC’s are constantly judging you as a founder, you should be equally mercurial in your judgement of them I totally understand if you’re an early-stage founder to have to be patient with low quality investors, but I feel like startups that are more established should call them out on their bullshit!
86
87
29
1,457
136,682
Ryan ᵐᶠᵉʳ 🦄d/acc retweeted
大家好,刚来到推特,自我介绍一下吧: 我是加辰,今年19岁,在运营一个装置艺术工作室 《 元素实验室 》 我滴绰号:激光少年🌈 自学|3D打印|小红书元素实验室|装置艺术| VR游戏|剑术|弓箭|举铁 小艺术家 ! 🖼 我眼里的我 追求刺激的野生工程师,热爱运动,音乐成瘾 通常情况下对科技和艺术有着爆棚的好奇心 喜欢挑战权威 下面是我在飞书上完整的个人说明书: fcno6t1gx32s.feishu.cn/wiki/… 下面🔗是我小工作室的3D模型: sry被子没叠,我懒 点链接进入3D虚拟空间kiriengine.app/share/3dgs?ta… 希望能够在推特遇到可以一起玩一起搞事情的朋友🫰 下面有一些我做的装置《声音投影仪》&《可穿戴光路》 我的小红书号:182543474
1
1
2
379
Ryan ᵐᶠᵉʳ 🦄d/acc retweeted
This is what would-be founders should be doing when they're young, not bogus startup pitching competitions.
This Project got me an interview at Google
130
447
39
8,160
697,531
Ryan ᵐᶠᵉʳ 🦄d/acc retweeted
持续制作这个系列! 卡比兽:你再不来,我要下雪了!
163
9,156
254
15,507
1,395,447
Ryan ᵐᶠᵉʳ 🦄d/acc retweeted
大家好!我是 Foresight Ventures 的合伙人兼北美区负责人。今天我们纽约办公室正式落成,借这个大喜的日子,想和大家分享下一路上的故事和感悟。放心,我不是来念官腔或者灌鸡汤的。 先自我介绍一下:
我是 Foresight 最年轻的女性投资合伙人。(“最年轻”这个头衔估计用不了几年了,趁还能蹭一下热度,先小骄傲一把吧。)VC 行业本来就不算女性扎堆的领域,在 Crypto 这更是稀有物种级别的存在。一路走到现在,说不辛苦是假的,一路上也少不了运气加持和贵人的帮助。 一路上我听到过很多好玩的评价。 从最开始的 “Foresight 真会招美女……”(嗯,谢谢夸奖,我只听进去了前半部分), 后来变成“我读过你写的研报,那是我入行启蒙之作!” ; 从“女生拿 deal 肯定比男生容易吧”,到现在 “听说你早就投中了某项目,能不能拉个群帮我认识下创始人?”  ;从“Foresight 是那个 Foresight News 的投资部门吗?”到 “我当然知道 Foresight Ventures,最近好多优质项目 captable 上都有你们。” 不过 Foresight 的起点确实不算“讨喜”。 2021 年底,我们成立在了熊市的初期,时机可以说差到不能再差了。我那时候刚从一家美元基金跳槽过来,心想着 Web3 大展拳脚,结果直接被现实按在地上教育。 我入职后接的第一个任务是写 Terra 的研究报告,连续熬了好几个通宵才搞定。结果 Terra 暴雷了,合作写文章的 Terra Foundation 研究员跑路了,接着 FTX 也爆了……不到三个月,我学会了一个真理:never say never。 熊市再难,也不能摆烂。
作为新人,我们深知“天上不会掉好 deal”,只能加倍努力。不够聪明怎么办?拼命补课!没有资源怎么办?拼命 networking!Conference 一个都不能落,生怕错过任何一个好项目或者优秀的创始人。 三年时间转眼过去,Foresight 在行业里杀出了一条血路。现在我们是亚洲成长最快、最活跃的 Crypto 基金之一。(没错,“卷”是我们的标配,但也是成长的一部份) 加密这个行业,注定是属于年轻人的狂欢。Foresight 一直愿意为新的范式、新的叙事、新的团队买单,哪怕成功概率只有 1%。因为我们相信,试错和探索是通往颠覆性创新的唯一路径。想成为第一个吃螃蟹的人,就得先敢下手。 但说实话,做 VC 难免踩坑,尤其是早期的“高 FDV 低流通率”项目……(有些 captable 就像讲笑话一样,不交学费真不知道这些数字是怎么编出来的。)  虽然被社区批评“VC 就是割韭菜”的声音少不了,但我们也确实越走越清醒: 1. 别和社区对着干,水能载舟亦能覆舟。 想当镰刀的,往往自己会被割。 2. 少一点“Wen token”的浮躁,多一点“Wen revenue”的务实。 3. 创始人应该关注长期价值,而不是短期热度。 再来点哲学思考:
如果有一天 AI 真把人类取代了(以它现在写研报的速度,确实很有可能),唯一无法取代的就是创造力。基于这种信念,Foresight 在纽约设立了办公室,和被投企业 The Block 和 Morph 一起,为行业贡献一点力量。 说是“开了个办公室”有点谦虚了。我们在 世贸中心 77 层,没错,THE World Trade Center。地标满分,风景拉满的全球金融中心,站在窗边俯瞰曼哈顿的纸醉金迷,感觉下一秒你就能搞个改变世界的大项目。(当然也可能只是在想晚饭吃啥)。 说笑归说笑,Foresight 的确在认真思考:Crypto 如何真正改变世界? 我相信整个行业都在寻找那个“Crypto 圣杯”——一场能推动大规模应用的范式级创新,而不是只停留在链上的 Meme 狂欢。而这种创新,往往来自那些敢于突破传统的年轻人。 如果你是一个有趣的年轻人(重点是“有趣”),随时欢迎来我们纽约办公室坐坐!喝杯咖啡聊聊未来的无限可能,和 The Block 的研究员聊聊行业趋势、八卦一下热点,或者和 Morph 的开发者开个通宵 Hackathon——肝项目的快乐,只有经历过的人才懂! 不在纽约也没关系,随时可以通过 Twitter 或邮箱联系我:[email protected]。 我们不想只当一家平庸 VC,我们更希望成为连接东西方社区的桥梁,为创新项目提供冷启动的环境与支持——哪怕只是一点点的助力,也能让好创意跑得更远。我们也意识到东西方 Crypto 社区在叙事和打法上有着显著的差异。Foresight 正在撰写一份关于亚洲市场 GTM 策略的年度报告,为行业提供更多insight。欢迎大家多多关注我们的官网! 保持期待,保持热情,一切才刚刚开始!🚀
207
34
10
706
210,106
Ryan ᵐᶠᵉʳ 🦄d/acc retweeted
在小说《巴别塔》中,“刻银术”是一种神奇的魔法,能够捕捉不同语言间最细微的差异,连接彼此难以相通的心灵。今天的 AI 技术,正在成为现实世界中的“刻银术”——它不仅是一种工具,更是一座座沟通的桥梁,让每一种独特的生命体验都能找到自己的声音。 这是一个关于技术的故事,但更是一个关于理解与包容的寓言。在规模与标准化主导的时代之后,我们也许终于找到了一种更温和的愿景。 这也是四十岁生日之际,我写给未来的一封信。
1
1
10
2,063
Ryan ᵐᶠᵉʳ 🦄d/acc retweeted
50
81
14
1,433
97,777
Ryan ᵐᶠᵉʳ 🦄d/acc retweeted
OpenAI Strawberry (o1) is out! We are finally seeing the paradigm of inference-time scaling popularized and deployed in production. As Sutton said in the Bitter Lesson, there're only 2 techniques that scale indefinitely with compute: learning & search. It's time to shift focus to the latter. 1. You don't need a huge model to perform reasoning. Lots of parameters are dedicated to memorizing facts, in order to perform well in benchmarks like trivia QA. It is possible to factor out reasoning from knowledge, i.e. a small "reasoning core" that knows how to call tools like browser and code verifier. Pre-training compute may be decreased. 2. A huge amount of compute is shifted to serving inference instead of pre/post-training. LLMs are text-based simulators. By rolling out many possible strategies and scenarios in the simulator, the model will eventually converge to good solutions. The process is a well-studied problem like AlphaGo's monte carlo tree search (MCTS). 3. OpenAI must have figured out the inference scaling law a long time ago, which academia is just recently discovering. Two papers came out on Arxiv a week apart last month: - Large Language Monkeys: Scaling Inference Compute with Repeated Sampling. Brown et al. finds that DeepSeek-Coder increases from 15.9% with one sample to 56% with 250 samples on SWE-Bench, beating Sonnet-3.5. - Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters. Snell et al. finds that PaLM 2-S beats a 14x larger model on MATH with test-time search. 4. Productionizing o1 is much harder than nailing the academic benchmarks. For reasoning problems in the wild, how to decide when to stop searching? What's the reward function? Success criterion? When to call tools like code interpreter in the loop? How to factor in the compute cost of those CPU processes? Their research post didn't share much. 5. Strawberry easily becomes a data flywheel. If the answer is correct, the entire search trace becomes a mini dataset of training examples, which contain both positive and negative rewards. This in turn improves the reasoning core for future versions of GPT, similar to how AlphaGo’s value network — used to evaluate quality of each board position — improves as MCTS generates more and more refined training data.
130
1,078
211
6,039
802,136