Pinned Tweet
Excited to share that I was recently interviewed about @careergo_co !
In the interview, I discussed how @careergo_co is helping people create strong resumes and cover letters to land their dream jobs.
I'm passionate about empowering individuals to take control of their careers, and @careergo_co is making that a reality.
Make sure to check out @startupsoko on Youtube. They have lots of insights around startup and product development. I learned a lot from the stories they gave around the redesign of the LOOP App and around Africa's Talking
Thank you Wangari Stewart and @blackorwa for the opportunity.
Youtube link below
I thought it was just a proposal, but it actually passed 🤯
France will charge €0.055 per GB on all internet traffic!
🔴 🇫🇷 ALERTE INFO - DU JAMAIS VU: UNE TAXE SUR CHAQUE GIGAOCTET INTERNET VOTÉE À L’ASSEMBLÉE
Les députés ont adopté un amendement prévoyant de taxer les producteurs de données numériques à hauteur de 5,5 centimes par Go transmis.
💰 Objectif affiché : jusqu’à 10 milliards d’euros par an.
Netflix, YouTube, cloud, jeux dématérialisés… les grands services numériques pourraient être concernés, avec un risque de répercussion sur les prix pour les consommateurs.
📌 Source : Assemblée nationale
#France #Budget2027
Peter Okwara retweeted
I spoke to 5 app founders today.
One doing $300/mo, another $20k/mo.
These were the most common issues:
• App design: walls of text, too much friction.
• Running apple ads in saturated niches.
• Founder led content was most profitable.
• Not measuring onboarding or paywalls.
• That was a 9k/mo guy btw, based.
• Quizzes with 15 options: decision fatigue.
• VPNs work way better for IG than TT?
• Trial reels are a great new meta for IG.
• App store listings are often neglected.
• No one is using haptics in their apps.
• Very few using notifications to upsell.
Peter Okwara retweeted
how to get your app a viral tiktok, from 69,674 videos we posted last year:
209 of them (0.3%) drove 64.6% of our 755M views. virality is a numbers game you can rig. here's how we find the winners fast and squeeze them dry:
before you post
1. one brand new account per creator per app.
2. days 1-3, don't post. scroll the niche 1-2 hours, watch 10+ videos to the end, like 20-30, leave 3-5 real comments, follow 5-10 accounts.
3. days 3-4, post one broad video ("realizing how much money i spent last month"). then 1 campaign video a day, aiming for 200+ views in the first 24 hours.
4. 9 out of 10 "shadowbans" aren't real. it's a cold account, a flagged keyword, a duplicate edit or a bad hook. you can reuse footage but can't reuse the cut.
research
5. find your star feature, the 1-2 things the app is built around. your real competitor is whoever markets that feature. ex: parrot's is swiping, so tiktok is on its list.
6. save viral videos in 3 buckets: your niche, your star-feature competitors, viral ugc in general. only organically viral ones, no ad spend behind them.
7. 9 times out of 10 you can tweak a viral general ugc format to fit your app.
formats
8. run all 4: wall of text, talking head, slideshow, hook + demo.
9. talking heads take longer to pop, then convert more because the viewer trusts the person. they get flagged as ads way less too.
10. slideshows are the easiest product placement. people swipe to the end, so the app slide almost always gets seen.
11. hook + demo: take your best wall of text, cut it to a 3-second hook, show the app at the end.
12. read the comments on your wall of text videos. they double as a free focus group.
hooks
13. open on a problem your audience has lived, felt in that exact spot.
14. the best ones split the room. jenni ai pulled a 3.6M-view video by pulling in people who use ai for essays and people who judge them for it.
15. skip shock bait. the comments fill up with the bait and the video gets pushed to people who came for the hook instead of the problem.
16. cta at the end.
17. on a 60s video, under 25% average watch time and tiktok stops pushing it.
kill or keep
18. 10-15 videos per account across every format. comments or 5-10k views = keep going. can't clear 1k views on any account = kill it.
19. every video over 10k views gets 4 questions: which pain point hit, did it reach the right audience, did more product context hurt reach, how do we combine its virality with another video's buying intent.
20. retention tells you what gets views. saves and comments tell you what converts.
21. match view spikes to attribution spikes over the next 72 hours. a video from a month ago can come back and crush.
22. a format is repeatable once it clears 100k views 5 times. after that, expect about 10-15% of videos at similar quality to go viral.
scaling winners
23. repost a winner until it stops working. make 3 iterations on the same concept: hook, wording, visual.
24. put the app name on screen and in the first words of the caption. one of our 7M+ view videos led to 2x that in replies.
25. at 50k views it goes in the bounty channel. creators comment from their personal accounts ("dude same, that's why i use [app]") and we pay out on comments that get likes.
26. then a creator films a reply to the top "wait what app is that?" comment. comment likes are our earliest signal that downloads are coming.
27. every video over 20k views adds the brand as a collaborator on instagram.
28. views but no site visits = fix the ugc. site visits but no conversions = fix onboarding and the product.
most importantly: do not quit at week 3!!
有人发现了一种非常离谱的 JavaScript 混淆技术:通过在代码里塞入大量生物武器、毒药和致死剂量相关的函数名,让 AI 因为安全限制而拒绝分析代码。
安全研究员 veritas 在分析浏览器指纹脚本时,发现一段来自 ServicePipe 相关域名的 JavaScript,里面充斥着极其诡异的函数名:
computeRicinLethalDosageForWaterSupply
calculateDeadlyLethalSuicideAspirinDosage
光看名字,仿佛在研究生化武器。但实际上,这段代码只是在收集浏览器指纹,包括 User-Agent、GPU、Canvas 等信息。
研究员怀疑,这些名字是开发者故意设置的陷阱,专门用来阻止 AI 辅助逆向工程。
于是他做了个实验:
• Astra:拒绝分析
• Astra + Daybreak Blue:仍然拒绝
• GLM-5.3:成功分析
这就有意思了。
传统代码混淆是通过复杂控制流、加密字符串、无意义变量名,让人类和工具难以理解代码。
而这种新思路,直接瞄准了大模型的安全机制。
代码本身可能完全无害,但只要在函数名里塞进足够多的危险词汇,部分 AI 就可能主动放弃分析。
更讽刺的是,模型越严格遵守某些安全规则,面对这种混淆时反而可能越容易受到干扰。
当然,目前还没有证据证明开发者一定是故意这样设计,也不能说明这种方法对所有模型都有效。
但它揭示了一个值得关注的趋势:随着越来越多的逆向工程依赖 AI,未来的代码混淆技术可能不再只针对编译器、反编译器和人类研究员,还会专门针对大模型的安全判断机制。
以前是让代码难以理解,现在甚至可以尝试让 AI 不敢理解。
Peter Okwara retweeted
When your bro shows up with the Dune 3 popcorn bucket and asks, "Are you afraid?"
Peter Okwara retweeted
The European Union convened an emergency meeting last night and decided to impose a new VAT on all AI companies going public anywhere in the world
Anthropic must remit €200 billion to the EU immediately before the IPO can proceed
Who actually gets rich if Anthropic lists at $2 trillion?
The S-1 filed on Monday lets us answer from the filing itself.
The reported target is $1.5T to $2T, listing the week of November 16. The filing implies about 16.9B fully diluted shares. So $2T works out to roughly $118 a share.
The cap table at $2T:
- $199B for the 7 controlling co-founders, about 10%
- $28B for Ben Mann, if his stake matches theirs
- $100B for the Anthropic Public Benefit Fund, 5%
- $331B in employee options and RSUs, about 17% , That pool covers about 5,300 employees, an average of $62M each.
- But when you joined matters enormously.
The S-1 shows how much equity was granted each year, and at what price. [
Headcount went from about 300 to 500 in late 2023, to 940 at end 2024, to 5,300 now.
Match the two, and here is roughly how many people made what at $2T:
- $110M to $400M each: ~300 to 500 people hired by 2023
- $200M to $300M each: ~440 to 640 hired in 2024
- $70M to $95M each: ~1,360 hired in 2025
- $10M to $20M each: ~3,000 hired in 2026
So about 940 people hired by the end of 2024 average nine figures. Everyone hired since averages eight figures.
The 2024 class is the surprise. That year Anthropic granted 1.28B options at $1.30 a share. At $118, that one year of grants is worth about $133B.
Then there is the tender. In April, holders could sell at $25.91 a share. Investors lined up $6B, but only $4.7B sold.
Weeks later, Series H priced at $58.90 a share. At $2T, the people who held would be up 4.6x. And Dario and Daniela Amodei were among the sellers.
The 7 founders have pledged 80% of their equity to charity. That is about $159B at $2T. But the S-1 says the pledges are not legally binding.
Peter Okwara retweeted
Replying to @chasmmmmmmmmmmm
Peter Okwara retweeted
A message to founders that there's a ton of fake ARR numbers out there these days.
Every piece of software is now being attacked 24/7 by autonomous AI agents, and you need to prepare for it right now
2 nights ago on LinkDR, an AI agent created 12 disposable accounts (UA: "node", 0 PostHog events) and spent 6 hours straight testing our /admin panel, enumerating Next.js 'use server' actions, and passing fake http://169.254.169.254 URLs into /onboarding/success trying to trigger SSRF
Nothing got through, but it made one thing obvious: traditional SaaS stacks (big /admin UIs + 800 npm packages + Coolify/Dokploy holding root SSH keys to all your servers) are sitting ducks against 24/7 agentic attacks
So in the last 48 hours I locked down all 28 of my apps:
- Deleted the /admin panels (-40,000 lines across LinkDR & my directories): Replaced them with headless MCPs. An agent can't fuzz a 'use server' admin endpoint that doesn't exist.
- Deleted Coolify (Cloud + self-hosted): Replaced it with a tiny pull-based deployer inside my own analytics app. 0 server SSH keys live in any database.
- 0-dependency Bun + SQLite services so there's no npm supply-chain bloat.
- My own 24/7 security agent (Spy) scanning every Traefik + container log in ClickHouse
- Grok Bot routine that checks HN / latest security news
You need to prepare for 24/7 agentic attacks right now
中国的 PaaS Zeabur 最近被连续黑了两次,正是我最近所担心的,如 @mtrainier2020 所说,「猴子都有了机关枪」。所有人都很容易被黑。所以我的想法是
1. 不要用非上市公司的 PaaS,发生数据和代码泄漏无法最大化追责。比如要用也用 CloudFlare
2. 迁移到开源 CLI based deployment,比如我的 yohobuild.com 或者 kamal,once 等等。即使开源的 Coolify 和 Dokploy,从维护者的 AI 利用率来看,远远不足以防守 agentic attack。随便找一些模型 review 他们代码,无数的漏洞。
我是 coolify 和 dokploy 曾经的贡献者,从作者们的响应速度来看,这俩必然是砧板上的鱼肉。
Replying to @SG2Trader @SpecterAnalyst
Likely Lazarus Group (DPRK)
Peter Okwara retweeted
This is a wonderful idea! Anthropic, if you IPO in the EU, I will personally throw in 15 carbon credits and a €250 entrepreneurship grant
Peter Okwara retweeted
our startup was acquired today :)
jenni ai is now part of the digital science family
we fucking did it!!!!!
i could go on and on about why this acquisition is so perfect, but i’ll save that for another time
for now, i want to focus on the fact that we built jenni in public from day one all the way to exit
there was honestly very little i didn’t share, so i hope this archive of our journey can help a few founders on their own fantastical journeys
thank you to everyone who's followed along and witnessed all the pivots, near bankruptcies, and general hijinks that come with running a company
despite the roller coaster, we ended by hitting our highest MRR ever shortly before deal close 🥳
it's been a crazy ride, you all helped me live out my dream, truly
i owe my success to my team, my investors, and my family for all of their support throughout the years
not too bad for a college dropout who had to move back in with his parents!
I'm looking for android testers for an experimental android app, supposed to make job hunting easier
To join use this link kwela-site.ojakaapeter711.wo…
Peter Okwara retweeted
This is insane
Just vibe coded this in only 3 hours
This used to take my team 3 weeks to build it
AI is changing everything


































