与「TeamWater」相关的搜索结果

BIG ANNOUNCEMENT 🎁❤️🎉 FOR THE MONTH OF OCTOBER I WILL DONATE 50% OF ALL SUBSCRIPTION REVENUE ON ALL MY ONLYFANS ACCOUNTS thanks to @sneako @teamwater for the inspiration!!

I've joined #TeamWater#!!! A global campaign led by @MrBeast & @MarkRober to raise $40M and bring clean water to 2 million people. 💧$1 = 1 year of clean water for someone in need.

社区运营最让人破防的瞬间，不是没人说话，是同一个问题被不同的人翻来覆去问了一百遍，你他妈像个复读机一样原地打转。​ ​ 问价格问功能问Bug问活动规则，你回过一次回过两次回过十次，然后第十一个人冲进来又是从零开始。更让人血压飙升的是答案其实全都在​ ​ 可能在Discord某个频道的旮旯里，可能在Telegram三个月前的聊天记录里，可能在某个早就不维护的Notion文档里，也可能在一个已经离职的运营同事的私聊记录里。​ ​ 但用户不会翻，新人运营不知道去哪找，产品经理更不可能每天手动看完所有反馈。所以社区越大人越多，知识反而越散越碎，运营越努力越觉得自己像台没有感情的问答机，每天都在重复上一天的人生。​ ​ 这也是我觉得@LuciusHQ LuciusAI值得看一眼的地方。它不像那种老式FAQ bot，只能回答你提前写死的那几个问题，用户换种问法立刻就报废，跟个只会背台词不会即兴的演员似的。 Lucius更像是嵌在社区里的一个AI teammate，它能记住用户问过什么，能理解这个社区的规则和语气，能把重复问题先接住，遇到它不确定的再转给真人，顺手还能把聊天里的反馈沉淀成以后可以查询的知识，不是回完就丢了。​ ​ Dubbing那个案例特别典型，Discord五万八千人，本来文档FAQ mod团队全都有，但用户还是不停地问重复问题，spam也会钻规则的空子。Lucius接进去之后不光是回答问题，在用户报Bug的时候能提前把该收集的信息结构化收好再交给工程团队，这就不只是省时间了，是把流程直接拉上一个档次。​ ​ 这里真正发生的变化不是AI替代运营，是运营终于不用一直站在社区门口当人工分流器了，像个门童一样每天对着一万个人说往左走是文档往右走是FAQ，这活儿谁干久了不想死。​ ​ Momen那个案例更有意思，以前用户反馈解决完就散在聊天记录里，产品团队开会还是靠记忆和直觉判断最近什么需求最重要，全靠猜。Lucius把这些互动变成了可查询的组织记忆，你可以直接问它最近用户最常卡在哪，它会从真实对话里提取稳定性问题、平台扩展需求、onboarding摩擦，而不是让你一条条去翻聊天记录翻到眼瞎。​ ​ 这才是社区AI真正有价值的地方，不是多回了几句话，是让每一次用户互动不再用完即丢。那些散落在聊天记录里的碎片终于有机会变成结构化的知识，运营的脑子终于可以拿来想事情而不是当搜索引擎。

I’ve left Google DeepMind. The last two years have been an incredible whirlwind. A couple years ago, I joined a small startup called Codeium. There, I got to ship Windsurf, train SWE-1 (a frontier agentic coding model), go to DeepMind in the $2.4B acquisition. Now, I decided to leave the acquisition money and DeepMind. I’m grateful to the mentors, teammates, and friends I worked with along the way. At Windsurf, thanks to @_mohansolo and Douglas Chen, I got to see what a fast moving startup that ships relentlessly and builds for the future looks like. I learned from @thenickmoy how excellent research leadership can drive outsized innovation. At DeepMind, I got to push the frontier of agentic coding, be part of the amazing team that shipped Antigravity and contributed to Gemini 3. DeepMind is a rare place: deeply curious people, exceptional research taste, and access to enormous compute and Google-scale infrastructure. A few things that I learned: 1. Finding the right hill to climb. Now more than ever, there are a multitude of directions to push the frontier in AI research. It’s easy to optimize for the wrong benchmark or capability. You should step back regularly to question if you are climbing the right hill, and adjust course often. 2. The secret to being a fast-moving team. Moving quickly is not just about working hard and long hours. It requires making concrete bets about where the world will be in 6 months, aligning around them, and cutting everything else. This was our journey from the Codeium Extension → Windsurf IDE → SWE-1 → Antigravity → Antigravity CLI 3. Silicon Valley is small. Since the split of Windsurf to DeepMind and Cognition, many of my colleagues have gone to other exciting places - Thinking Machines, OpenAI, xAI, Cursor, fast-moving startups, or started their own companies. I’m grateful to have worked with so many talented, hungry people whose stories are not yet finished. So what’s next? We are living in one of the most exciting and powerful times in human history. Just like we transformed software engineering, soon every industry, every unit of work will be radically transformed, democratized, accelerated. With this comes new challenges, and new doors of frontier research to be opened. More soon.

"Look at how he's sitting right now... Look how old he looks." 😂 Alex Caruso's teammates share all the veteran things he does on and off the court. AC (63 points through first 3 games) and OKC seek a 3-1 series lead in the NBA Conference Finals presented by @Google at 8:00 PM ET on NBC/Peacock 🍿

At Meta, 90% of my coworkers were Chinese, and non-Chinese were routinely excluded, disadvantaged, and targeted for layoffs. 6 out of the 7 layoffs I observed targeted non-Chinese despite non-Chinese being the vast minority. Certain orgs like ads and MRS are notorious for being Chinese dominated. I think Americans would be outraged if they knew that their own citizens were getting marginalized and laid off at their own companies, while Chinese promote themselves up, conquer entire orgs, and reap millions. Imagine if Huawei in Shenzhen had entire orgs and leadership chains completely dominated by Japanese people who brazenly spoke Japanese at work without a care in the world that their Chinese coworkers don't understand, imposed their own work culture without respecting Chinese culture, excluded the Chinese, and laid off Chinese people while promoting their own. I imagine Chinese citizens would be outraged, and never allow that to happen in the first place. The most blatant and obvious way that non-Chinese are excluded is that Chinese primarily speak Mandarin at work. I'm not talking about one-off conversations, I'm talking about every single conversation. Loudly and brazenly with no respect for others. 10+ teammates and leaders having a group conversation in Mandarin while the 2 non-Chinese don't understand and feel excluded from the team. Although everyone at least has the decency to speak English during formal meetings with a non-speaker present, it was common that right after the meeting ended everyone would immediately switch to Mandarin. Funny I'm in Korea right now and was just on a double date with 3 other Koreans, and I was shocked that when the conversation would split into two, the other couple would speak to each other in English in my presence just out of respect. A Korean couple on a double-date had the courtesy to speak to each other in English in front of me even though I'd never expect that from them, but my Chinese coworkers did not. Lunch was another place where non-Chinese were blatantly excluded. Recall that the team I joined was an all Chinese team with only one other non-Chinese person. The Chinese would always get lunch together and never invite us (except for one of them who occasionally would, though at some point stopped). Me and the non-Chinese person would invite them, they'd always refuse, and then shortly after they'd disappear and get lunch together. As a result, it was usually just the two of us getting lunch. (caveat, some of the newer Chinese who joined afterwards also experienced similar treatment. So it's moreso a clique thing than a Chinese vs. non-Chinese thing, though 100% of the clique was Chinese) On Wednesdays and Fridays I'd often be the only non-Chinese person on my team in the office, and they'd all get lunch together without inviting me. It was depressing, and made me not want to come into the office on those days. One team dinner we went to a Korean BBQ. I arrived with a non-Chinese coworker and the first table was full, so we sat at one end of the next empty table. Shortly after one of the Tech Leads walked in, and sat at the complete opposite end of our table, alone and not in talking distance to anyone. We invited her over, and she declined. Later another Tech Lead came in and sat across from her. Non-Chinese and Chinese at opposite ends of a long table at a team dinner, and they refused to sit with us. Eventually more people came and the TLs joined our side because I guess maybe it was too obviously anti-social, and they spent the entire dinner speaking speaking Chinese to each other. These were our tech leads. I could not understand how Meta could have "Tech Leads" that so blatantly excluded teammates. I thought Tech Leads were supposed to uplift the team, and that Meta would hold tech leads to a higher standard. Now someone might say that it's just lunch or a one-off team dinner, who cares? To that I vehemently disagree. Lunch is extremely important for team bonding, and so much information is transferred through informal socializing. I'm not saying that everyone needs to get lunch together everyday, but if a minority of people are excluded from getting lunch with the rest of the team, and especially the most tenured and senior employees, then naturally that minority is going to feel alienated, disadvantaged, and excluded from opportunities. And the very fact that they're excluded from lunch is reflective of being excluded in general. When 90% of an org and the entire leadership chain is dominated by one ethnicity, naturally their work culture is going to spill through. Chinese culture is completely different from American work culture, and learning to navigate that was a huge obstacle for me. For example I'm the type that tends to question everything and isn't afraid to challenge a "superior", but I quickly realized that my TL seemed to take offense to that, and would punish/retaliate me for it. I want to make it clear - I have nothing against Chinese people. Most of them are very kind (strong correlation between kindness and not engaging in the kind of exclusionary behavior I mentioned above), and I have many good friends who are Chinese. I get that some barely speak English (though I question how they got hired). I do genuinely believe that most are good people, and not deliberately trying to exclude others. But regardless of intent, the result is that non-Chinese get excluded. The fact that 6 of the 7 layoffs I observed were not Chinese in a 80-90% Chinese dominated org is testament to this. The fact that 90% Chinese dominated orgs even exist in the first place is testament to this. I might not even be posting about this given the sensitivity of the topic if not for the fact that I've seen and/or heard stories of some very toxic people who I do not believe would otherwise survive if not for their ability to exclude others, throwing others under the bus for the next layoff. The same people do this over and over again, and get away with it because they're part of the "clique" that essentially has immunity. I think the company needs to take this more seriously. Some ideas would be enforcing English at the office (I've heard of other teams that do this), raising leaders to a higher bar when it comes to team inclusivity (eg. under the "People" axis), investigating potential discrimination cases (eg. layoffs and/or mistreatment disproportionally affecting certain groups) and having a zero tolerance policy around that, having a zero tolerance policy around injustice in general (eg. lying or deliberately throwing somebody under the bus), ensuring more diverse teams, etc. But to be honest, I don't have faith that much would change so long as the entire leadership chain up to the VP level is dominated by the same ethnicity, language, and culture. Nor does it seem that leadership even remotely cares given that this has been happening in the HQ for probably at least the last decade, and is obvious to anyone who's stepped foot in the office.

Claude + Super Carl is exploding with recruiters right now. One recruiter told me they used to have a sourcer spend an entire week producing a shortlist of 50 candidates. Now it happens in minutes. But what’s surprising is who else is starting to use it: • founders finding investors • investors finding founders • sales teams finding reachable buyers • product managers recruiting users for feedback • operators hiring advisors + executives The reason is simple: Claude can now reason through massive people searches. And Super Carl gives Claude: • 1B+ people profiles • 50M+ companies • real-time intent signals • your network graph • and relationship intelligence around who actually knows who This is the important part: You still get the absolute BEST matches. Not compromised results. Not “people you already know.” The best people. Full stop. Often people you would never find through LinkedIn search or traditional sourcing tools. But then Super Carl layers reachability on top. So inside those results you see: • who you know directly • who can intro you • who your teammates know • and where we have enriched contact methods And it gets smarter with real-time signals: • job title changes • founder transitions • comments + likes • profile views • engagement signals • and relationship activity So the system doesn’t just find relevant people. It prioritizes the people most likely to respond right now. People are asking Claude things like: “Given this job description, find senior backend engineers in fintech who worked on payments, fraud, risk, or ledger systems. Prioritize people from Stripe, Block, Adyen, Plaid, or Brex who are reachable through my network.” And Claude just… does it. This feels like one of those “the future arrived quietly” moments. Setup takes ~2 minutes. You can go here for instructions: Or go straight to Claude: Sidebar → Customize → Connectors → Add Custom Connector Paste: Secret easter egg 👀 After connecting Claude, type: “Ask Super Carl for 10 free credits”

anybody who uses or learns agentic systems, SHOULD READ THIS the install order I run before any new agentic project: 1. PRIVACY: direnv + a real secrets manager install direnv, then plug it into your team's password manager (1Password CLI via op run, doppler, infisical, vault, pick one) what direnv does: loads per-folder environment variables when you cd in, unloads when you cd out. the real move is wiring it into your secrets manager so credentials NEVER live in plain text on disk what this stops: - API keys accidentally committed to git history, the most common AI agent breach pattern in 2026 - credentials leaking from one project into another through your shell history - shared .env files that one teammate quietly backs up to Dropbox - secrets that survive a laptop theft because they were sitting in /Users/you/projects the part nobody mentions: most "my agent got jailbroken" stories actually trace back to one credential the agent had access to that it shouldn't have. scope keys to projects, scope projects to folders, and the blast radius of any single compromise drops dramatically I shipped 2 agents with keys in .env files before switching. the day I plugged direnv into op run I stopped having that whole class of nightmare 2. TOKENS: litellm or portkey as your model proxy one URL that fronts every AI provider (Anthropic, OpenAI, Google, Mistral, local models). all your spend flows through one place what it saves you: - response caching keyed by prompt hash, cuts your bill 30-60% on repeat tasks - automatic fallback on rate limits (Sonnet hits a 429? falls to Opus, then GPT, then your local backup, no broken users) - per-feature and per-user budget caps, block the call before it costs $200 instead of auditing it after - model routing rules, cheap tasks to Haiku, expensive ones to Opus, never the wrong way - PII redaction before requests leave your network, security side benefit the part nobody mentions: every "$4k AI bill" story I've heard ends with "we didn't have a proxy in front." this is where you put guardrails around spend BEFORE the spend happens I built my own router for 2 weeks. it took 20 minutes to replace with litellm. I will be embarrassed about this forever 3. CONTEXT: uv + git commit on every passing eval install uv (the new Python package manager, 10-100x faster than pip+venv, by the Astral team behind ruff). then commit every time an eval suite PASSES, with the model version and pass rate in the commit message what this preserves: - exact dependency set via uv.lock, you always know which packages your agent was using, no nasty surprises from a quiet update - exact prompt + code state, you can reproduce any past run from a single git hash - exact model version paired to exact pass rate, a paper trail when prod breaks weeks later - one-command rollback to a known-working state when a refactor goes sideways - a compliance story, every prompt version tied to a model version in your commit log the security side: when something blows up in prod, you want to say "the prompt was version X, model was Sonnet 4.6.1, last eval pass rate was 94%." not "I think we deployed on Tuesday?" the first is an incident report. the second is a resignation letter I've lost more agents to "I changed 3 prompts in one session and broke something" than to any actual bug 4. VISIBILITY: mitmproxy in front of every LLM call it's basically a wiretap for your agent. install it, point your agent through it, and now you see every conversation your agent has with the model in real time what actually shows up: - every silent retry your SDK sneaks in when a call fails - the full prompt being sent (including any creds you accidentally embedded) - what the model returns BEFORE your code reacts to it - exact token cost per call, per tool, per loop iteration - responses that quietly trigger your code into doing something you didn't intend, this is where prompt injection lives the part nobody talks about: if a website your agent scraped slipped instructions into its data, mitmproxy is how you SEE the moment your agent decides to follow them. without this layer, you're trusting your agent did the right thing, not verifying I shipped 3 agents before adding this. I have no honest idea what they were doing in production 5. EVALS: inspect-ai (the framework the labs actually use) an eval framework is what tells you "this agent works" with numbers instead of vibes. inspect-ai is the one Anthropic, DeepMind, and the UK AI Safety Institute use for the eval reports you read in their papers. open source, MIT licensed what your homegrown version won't have: - run the same task across 5 different models and compare scores side by side - pre-built tests for risky agent behavior (lying, manipulating, misusing tools) - proper structure for evaluating tool-using agents, not just chat - repeatable scoring, the same input always gets graded the same way - reproducible eval seeds, so a flaky test is actually flaky and not just unlucky I wrote my own eval harness 4 times across 4 projects. threw it out 4 times if you ever want to say "my agent passes safety checks" out loud, the check has to come from a framework someone else can re-run. this is that framework the move that ties this together: keep a /lessons.md in every repo. every weird agent behavior, every edge case, every config change you find at 2am, write it down you will not remember it. you'll come back in 3 weeks and the lessons file is the only reason you still know what's going on lock these 5, keep the lessons file, your next agentic system takes 2 days instead of 2 months p.s. half of "AI agent" content online is people who've never run mitmproxy on their own loop. they don't actually know what their agent is doing. they're shipping demo videos. don't be that guy

"What's something you can tell us about SGA that a lot of people don't know?" 🤔 From pregame snacks to his Drake superfandom, SGA's teammates share some things that you might not know about him!