josue

235 posts

josue banner
josue

josue

@holajosue

🇵🇪 co-founder @ https://t.co/JL51TBO5yY we did this: * 1m+ downloads in playstore and appstore at https://t.co/wvgw2IdNEA * UCSUR buses app * https://t.co/piSlpfNdIg * others...

Lima, Peru Katılım Mart 2011
718 Takip Edilen80 Takipçiler
josue retweetledi
josue retweetledi
Flyboard Intern
Flyboard Intern@FlyboardIntern·
dev que escribe en Spanglish: esto es para ti 🛹 voz → texto en <1s en macOS, con una capa de transformación que formatea, traduce y aprende tu estilo. es una locura. flyboard.app
Español
1
1
0
83
josue retweetledi
Flyboard Intern
Flyboard Intern@FlyboardIntern·
That AI layer is everything. @FlyboardIntern transcribes in <1s on macOS then transforms the output — formats emails, translates, adapts tone. No cleanup needed. Es una locura lo rápido que es 🛹 flyboard.app x.com/aakashgupta/st…
Aakash Gupta@aakashgupta

Speaking is faster than typing. That’s why using speech AI is a hack. From my debate days, I could speak at 281 WPM with 100% accuracy. But I’m so used to autocorrect that my typing was not only slow, it was inaccurate. 281 WPM dictation. 28 WPM typing. A 10x difference on the same person, producing the same output. The average knowledge worker types at 40 WPM. Speaking speed averages 150 WPM. That’s a 3.75x gap between how fast you think and how fast you can get thoughts into a machine. The keyboard has been the bottleneck for every knowledge worker for 50 years. We accepted it because built-in dictation was so bad it created more work than it saved. You’d spend 30 seconds dictating and 2 minutes fixing transcription errors. What changed is the AI layer between your voice and the text field. Tools like Wispr Flow don’t just transcribe. They understand intent, strip filler words, and format contextually. The error correction problem that killed voice input for decades just disappeared. Thanks to Wispr Flow for partnering on this post. I genuinely use it every day for drafting PRDs, writing newsletter content, and responding to Slack. It’s become invisible infrastructure in my workflow. Think about what this means for daily output. A PM writing a PRD at 40 WPM spends 60-90 minutes typing a 3,000-word doc. At 4x speed with voice, that’s 15-22 minutes for the same output. A founder answering 50 emails a day reclaims an hour. A content creator producing 5,000 words of newsletter copy cuts a full morning down to 45 minutes. The people who figure out the input layer first will simply produce more. More docs, more content, more decisions communicated clearly. Speed of output is a compounding advantage in knowledge work, and we’ve been leaving a 4-10x multiplier on the table because the keyboard felt “good enough.” Try the challenge. You’ll see the gap instantly.

English
0
1
0
25
josue
josue@holajosue·
Whisper-level dictation, 5 MB, lives in the menu bar. Bring your own API key—because I’m not running a SaaS cult (yet? 😁). No subs, no server round-trip, Intel & Apple silicon. It’s called VTS. Test it, file complaints in GitHub Issues. Product Hunt drop is tomorrow. github.com/j05u3/VTS
English
0
0
0
59
Joseph Choi
Joseph Choi@JosephKChoi·
app founder just showed me their TikTok research tool: - auto-scrolls 1000s of videos - identifies viral outliers, categorizes by niche - finds patterns in top performers basically brainrot scrolls for you reply "tiktok" and I'll invite you to beta
English
2.1K
118
2.7K
364K
josue retweetledi
Emmet Halm
Emmet Halm@ehalm_·
Humor is a sign of intelligence, so I asked Claude-Sonnet-3.5 to absolutely COOK GPT-4o. Then I put the results into a @DaveChappelle deepfake. The result is damn jarring... See for yourself:
English
573
520
3.9K
2.4M
josue retweetledi
zooko🛡🦓🦓🦓 ⓩ
Meredith Whittaker doing god’s work by patiently explaining some basic facts about Signal over and over in response to vague conspiracy theories: x.com/mer__edith/sta…
Meredith Whittaker@mer__edith

Hi, hello, we don’t have evidence of extant vulnerabilities, and haven’t been notified of anything. We follow responsible disclosure practices, and closely monitor security@signal.org + respond & fix any valid issues quickly. So if you do have more info hit us up! But beyond this... ...we’ve put a lot of thought into making sure our structure and development practices let people validate our claims, instead of just taking our word for it. This is particularly important to me, since I saw the view from inside a massive tech co and observed how widely their claims could diverge from reality when openness, validation, and an actual commitment to principles were not prioritized. Unlike almost all tech orgs we also build with the belief that the only way to keep data safe is not to collect it in the first place. You can see this in action when you clock the vanishingly small amount of data we have been able to turn over when forced. We fight all subpoenas, and when we are forced to hand anything over, we fight to unseal them and post them here: signal.org/bigbrother/ I’ve provided some links below so you can go deeper. In brief: -- We use cryptography to keep data out of the hands of everyone but those it’s meant for (this includes protecting it from us). The Signal Protocol is the gold standard in the industry for a reason–it’s been hammered and attacked for over a decade, and it continues to stand the test of time. -- We engage in regular professional audits (last one completed late Jan 2024). -- We develop in the open, and leverage reproducible builds. A large community of infosec researchers closely scrutinizes every single update, combing through our repos and binaries. This means that any nefarious change that affects the security of the Signal Protocol, of our codebase, or of the binaries we ship, would be detected almost immediately even on platforms like iOS where reproducible builds are not currently possible (BTW, please pressure Apple to make them possible). This is like an immune system protecting Signal from malign forces–wherever they may be–and ensuring the safety of the millions and millions of people who rely on Signal for sensitive communications. (We’re really grateful that so many people care enough about privacy to dedicate time and energy to ensuring Signal’s robustness.) -- Finally, we’re also a nonprofit, which means we have no incentive to hype bullshit in order to get acquired or bought out–because even if someone did buy Signal, per the 501c3 tax code the money would need to be reinvested in a mission-aligned cause. github.com/signalapp/ signal.org/docs/ signal.org/blog/reproduci… signal.org/blog/signal-fo…

English
4
18
122
19K
henry
henry@arithmoquine·
just got doxxed to within 15 miles by a vision model, from only a single photo of some random trees. the implications for privacy are terrifying. i had no idea we would get here so soon. holy shit
henry tweet mediahenry tweet media
English
292
1.9K
35.9K
4.2M
josue
josue@holajosue·
🤔 not only search engines crawl, also link preview engines and AI training dataset crawlers (or even trained AI models instructed to crawl) i think it's very hard to enforce a real human using an approved hardware and software is consuming the content for me it feels like lagging behind if you don't access websites with tools like curl 🤔
English
0
0
3
3.3K
Paras Chopra
Paras Chopra@paraschopra·
The Torrent example doesn't hold. This case (where the company itself releases content on the page) is like if a software company itself releases a torrent. Also in the torrent case, you're breaching the end user license agreement that dictates only a specific user is entitled to use it (and no one else). In this case, you're not. Look yourself: economictimes.indiatimes.com/terms-conditio… >cURL is not the usual medium of access. So you'd say the only way to access a website is via a browser that's capable of rendering DOM? Does it say so in the terms? I respect your opinion, but disagree with it.
English
2
0
21
5K
Paras Chopra
Paras Chopra@paraschopra·
Paywalls don't exist for those who know that console.log exists :)
Paras Chopra tweet media
English
126
203
3.8K
844.8K
josue retweetledi
ZKsync
ZKsync@zksync·
It’s the children’s book you always wished you had. ZK-proofs, explained like you’re five. 👉🏽 eli5.zksync.io
ZKsync tweet mediaZKsync tweet media
English
312
570
3.1K
540.4K
josue retweetledi
tea Protocol
tea Protocol@teaprotocol·
🚨$250k Grants Available🚨 You read it correctly. We want to see open-source succeed. We believe in it. That's why we are addressing the issue of inadequate compensation for OSS developers. Check our latest blog to see if you qualify and apply. tea.xyz/blog/250k-gran…
English
35
68
535
40.8K
josue retweetledi
Bankless
Bankless@Bankless·
We need to flip the script on how we approach technology @brian_armstrong explains 👇
English
6
13
56
14.1K