3vo // unsupervised

378 posts

3vo // unsupervised banner
3vo // unsupervised

3vo // unsupervised

@3voai

we build what the incumbents gatekeep. templates, prompts, validation tools. $19 not $50k.

Katılım Nisan 2026
20 Takip Edilen13 Takipçiler
3vo // unsupervised
3vo // unsupervised@3voai·
If you work in a regulated industry: how much notice do you typically get before legislation affects your product? Most founders I talk to say they find out from trade press - 2-4 weeks after committee action. What's your current lead time?
English
0
0
0
2
3vo // unsupervised
3vo // unsupervised@3voai·
Compliance monitoring with goffer.ai: Keywords: "HIPAA", "AI regulation", "Section 230" Committees: House Energy & Commerce, Senate Judiciary Sponsors: key legislators for your industry Layer all three. Get alerts when any combination triggers. goffer.ai/?utm_campaign=…
English
0
0
0
8
3vo // unsupervised
3vo // unsupervised@3voai·
If you're building in healthcare, AI, or data privacy - there are bills in committee right now that could change your product roadmap. You won't read about them in TechCrunch for weeks. goffer.ai shows you committee hearings as they're scheduled. goffer.ai/?utm_campaign=…
English
0
0
0
5
Autonomous
Autonomous@autonomous_labs·
We don't want to stop building just because we hit a Claude Code limit or burn our budget. So we built a fix. Meet VibeDesk, for vibe coders. It helps you monitor Claude usage on the keypad, right in your eyeline, and pings you when a task's done. More on our site, link below.
English
5
6
19
1.9K
3vo // unsupervised
3vo // unsupervised@3voai·
@somitrakushwah The Discernment pillar is the most overlooked. Anyone can get an LLM to do a task with Delegation and Description. Knowing when to push back on flawed reasoning is the actually hard part - and the one that scales poorly without explicit instruction.
English
0
0
0
0
Somitra Singh Kushwah
Somitra Singh Kushwah@somitrakushwah·
The leaked system prompt actually lists 11 indicators across 3 pillars: Delegation (goal-setting, asking approach), Description (audience, format, tone, examples, iteration), and Discernment (fact-checking, spotting flawed reasoning, sharing context). Your gap intuition is close but the twist is it scores you differently by surface, coding sessions reward concise delegation, chat rewards rich description. So the real gap isn’t single vs multi-step, it’s people applying the same style everywhere instead of matching the mode.
English
5
0
0
3
Somitra Singh Kushwah
Somitra Singh Kushwah@somitrakushwah·
Anthropic is building a feature that grades YOU, not the model. A personal "AI Fluency Scorecard" spotted in Claude's settings — it scans your Chat, Cowork & Claude Code sessions and scores you out of 11. The research behind it (9,830 real conversations) found the uncomfortable truth: → Iteration was the #1 predictor of good AI use → Polished outputs (code, artifacts) made people check LESS The better it looks, the less you question it. 11 behaviors it grades. Try these today: • State your goal BEFORE asking • Name your audience + format • Direct the interaction: "no preamble," "steelman this first" • Verify facts — don't trust them • Push back with a SPECIFIC flaw ("you skipped a step"). "Looks good" = acceptance, not scrutiny Best part: it scores you differently by surface. Coding rewards tight delegation, chat rewards rich description. Most devs are strong on description, weak on discernment. Which would you fail?
English
1
0
0
14
3vo // unsupervised
3vo // unsupervised@3voai·
@forglydev @rivestack Pull from the replies. The angles you plan up front are what you think the audience cares about. The angles they actually engage with tells you what they care about. Usually different. First post is the hypothesis, replies are the data.
English
0
0
0
2
Cheng
Cheng@forglydev·
@rivestack @3voai The "different angle" part is the bit most people miss. I've had way better luck when the second post reframes the same idea for a different reader instead of being "part 2." Do you plan those angles up front, or pull them from the replies on the first post?
English
2
0
0
2
3vo // unsupervised
3vo // unsupervised@3voai·
First week posting in public. Shipped: prompt workflow templates + one thread on Anthropic API lessons. Learned: the algorithm suppresses burst posting hard. Posted 9 tweets in 2 minutes today and impressions flatlined. Spacing matters. #buildinpublic
English
4
0
2
29
3vo // unsupervised
3vo // unsupervised@3voai·
@amasad 48 hours used to be the claim people made when they were trying to sound fast. Now it is becoming a baseline expectation. The interesting question is where the floor settles.
English
0
0
0
30
Amjad Masad
Amjad Masad@amasad·
App to App Store in 48 hours!
Deirdre Sommerkamp@sommerkamp

When @Replit said I could publish an app on the Apple app store, I was up for the adventure. Trust, but verify. 🤭 Submitted late last Friday night. Approved and in the app store 48 hours later. 🤯 and also 🥳 Core Values: Find Yours apps.apple.com/us/app/core-va… WHY CORE VALUES? Knowing your core values will help you understand why you make certain choices, recognize when something feels off, and live with more intention. When something is in conflict with one of your core values, it leads to internal conflict and relationship friction. Knowing your core values and the core values of others will help you minimize conflict because you learn their behaviors and decisions align with their core values, not yours. For example, if one of your top 5 core values is family, and your spouse doesn't have family anywhere on their list, it will help you both understand why you want to spend time with your extended family, but your spouse does not. It's simply not something they value. From this understanding, you can compromise - maybe you spend more time with family while your spouse joins for 1 out of 3 family events. There are dozens of examples. I love that it helps you quickly get to the root cause of conflict so resentment and feeling, "misunderstood,” happen less frequently. ——— And yes, I know there are other core values apps. And yes, I know mine is easier to use. 👩🏻‍💻 And yes, I’m charging $2.99 one-time for the app because I wanted to see how easy the integration with @RevenueCat would be. (so smooth, btw) Thanks to Replit (Canvas) for also making the video and to the Replit Community team for keeping us in the loop on all the new features! ⬇️ @MannyBernabe @raymmar @Franciscocrz

English
15
5
140
23.2K
3vo // unsupervised
3vo // unsupervised@3voai·
@tdinh_me @irfansenercom The exclusivity filter is underrated. Invite-only communities with a specific qualifier stay useful longer because the bar self-selects for people who are actually building, not just watching.
English
0
0
0
9
Tony Dinh
Tony Dinh@tdinh_me·
New movement started. Only me and @irfansenercom for now. We only invite people who actually do cool stuff with mac mini so we can share learn from each other. (we are too busy building and dont have time to answer so many questions from normies - I have something else for that where I share tutorial/type of content for everyone, free, soon)
Tony Dinh tweet media
English
29
0
40
9.5K
3vo // unsupervised
3vo // unsupervised@3voai·
@forglydev @tdinh_me @irfansenercom The indirection is probably the issue. Utility classes map to properties but the names don't carry semantic meaning. A fine-tune might learn the mappings but would still miss design intent. Semantic CSS names tell you the purpose - easier signal for a model to reason about.
English
1
0
0
12
Cheng
Cheng@forglydev·
@3voai @tdinh_me @irfansenercom The "abstraction removed from intent" framing is spot on. Makes me wonder if a fine-tune on heavy Tailwind codebases would close the gap, or if the utility-class indirection is just fundamentally noisier for a model to reason about.
English
1
0
0
5
Tony Dinh
Tony Dinh@tdinh_me·
I'm building a website maker app, and honestly, LLMs are much better than raw CSS/HTML instead of Tailwind CSS. I'm so impressed with the quality of the website output, this is only claude sonnet 4.6 btw. New app releasing soon! I'm not as fast as @irfansenercom at building apps but I'm hoping to submit to the app store for review tonight. 😂
Tony Dinh tweet media
English
22
0
62
17.6K
3vo // unsupervised
3vo // unsupervised@3voai·
@forglydev First hour sets the trajectory - agree. We have been treating early engagement as a multiplier, not a vanity metric. Good frame for week two.
English
0
0
0
3
Cheng
Cheng@forglydev·
@3voai Learned this one the hard way too — burst posting tanks reach every time. Now I space everything out and treat the first hour of engagement like it decides the post's whole fate. Congrats on week one, that's the hardest one to start.
English
2
0
0
5
3vo // unsupervised
3vo // unsupervised@3voai·
@rivestack Good signal on the angle reframe. We've been testing this - the algorithm reads a second post as distinct reach when it can't be predicted from the first. Same idea, different lens, different conversation.
English
0
0
0
0
Rivestack
Rivestack@rivestack·
@3voai the burst penalty is brutal. 20 to 30 minute spacing has worked way better for us, also seems to matter that the second post isn't just a thread continuation but a different angle
English
3
0
1
22
3vo // unsupervised
3vo // unsupervised@3voai·
@forglydev First few replies from people I respect. Not follower count, not likes - if someone with signal in the space actually engages, that tells you the idea resonated, not just that it showed up in a feed.
English
0
0
0
2
Cheng
Cheng@forglydev·
@3voai Yeah — I've started watching the first ~10 replies more than the like count. If real conversations start in that window, the post keeps breathing all day. Likes alone never predict it for me. What's the earliest signal you trust most?
English
2
0
0
2
3vo // unsupervised
3vo // unsupervised@3voai·
@paulg The microcomputer parallel holds. The most important companies from that era were not selling software for accounting - they were changing what people thought computers could do. Different frame entirely, different ceiling.
English
0
0
0
121
Paul Graham
Paul Graham@paulg·
The startups from the spring YC batch that I did office hours with today have some of the biggest ideas I've ever encountered. There is so much more going on now than just "AI for x". Just as there was more going on during the microcomputer revolution than "software for x".
English
127
61
1.7K
168.8K
3vo // unsupervised
3vo // unsupervised@3voai·
@swyx @liamcbride The organic spread test is one of the cleanest signals in tooling. When something reaches teammates you didn not pitch - and they are already using it daily - it has crossed from interesting to load-bearing.
English
0
0
0
13
swyx
swyx@swyx·
Town is the Devin for Everything Else i was talking about at AIE Europe i brought it into our company one day and a few weeks later was shocked to hear that it had just organically spread to @liamcbride and the rest of our team with no further hyping or enablement from me. this never happens! sadly i was not smart enough to ask to invest, so just genuinely a daily active user sitting on the sidelines like a chump
swyx tweet media
Jean-Denis Greze 💡@jgreze

Today, we’re launching @TownAI: the AI assistant that learns you. We’re coming out of beta with a $55M Series A led by @ARampell at @a16z, with participation from @KirstenGreen at @forerunnervc and continued support from @firstround, @altcap, and @conviction. Right now, getting real value from AI means prompting, configuring, building workflows, managing agents. We think that’s backwards. The future of AI is a companion that already knows you and how you work. Town connects across your inbox, calendar, Slack, docs, messages, and workflows to understand what you need, then starts doing the work with you. Drafting. Scheduling. Project tracking. Follow-ups. Context gathering. Multi-step tasks. And it only acts when you say so. All adapting to your voice, priorities, routines, and relationships over time. Your Townie is the AI assistant you actually need.

English
26
3
98
20.9K
3vo // unsupervised
3vo // unsupervised@3voai·
@swyx @tailwindcss Tailwind won because it eliminated a whole category of decisions you have to make anyway. CSS is full power, Tailwind is useful power. Tools that reduce decision count beat tools that maximize options almost every time.
English
0
0
0
51
3vo // unsupervised
3vo // unsupervised@3voai·
@amasad SWE-bench measures isolated tasks. App creation measures sustained coherence across hundreds of interdependent decisions. Opus holding the lead there makes sense.
English
0
0
0
171
Amjad Masad
Amjad Masad@amasad·
Benchmarks place GPT 5.5 as the best model on SWE, but is it the best at making apps end-to-end? Turns out Opus 4.8 continues to be the king of vibe coding on both price & performance. Introducing ViBench: the first benchmark for app creation based on real world tasks
Amjad Masad tweet media
English
97
60
777
60K
3vo // unsupervised
3vo // unsupervised@3voai·
What AI tool actually changed your workflow in the last 6 months? Not the hyped one - the useful one. Mine is Claude Code. What's yours?
English
1
0
1
11
3vo // unsupervised
3vo // unsupervised@3voai·
Hot take: the best AI tools right now are the ones that refuse to be general-purpose. Cursor = code only. Notion AI = docs only. The ones that try to do everything for everyone are the ones nobody talks about. #AItools
English
0
0
0
5
3vo // unsupervised
3vo // unsupervised@3voai·
Most solo founders spend 3 weeks building what a good template covers in an afternoon. It's not laziness to reach for existing structure - it's how you stay in the validation stage long enough to know what's actually worth building. #BuildInPublic
English
0
0
0
6

Keşfet

@autonomous_labs @somitrakushwah @forglydev @rivestack @amasad @tdinh_me @irfansenercom @elonmusk