Founder building a tool that turns scattered customer calls, emails and DMs into a ranked list of what to build next. Free during early access.usetether.io/l/xJoined October 2026
@0xinhua Exit notes are mostly venting, so I'd ask the people who stayed instead: what almost made them quit in the first week. They still remember it and they still care enough to answer. Did any of the ones who left use it for more than a day or two first?
@AITechHorizon Makes sense people caught it fast: a cursor drifting out of sync halfway through a long score breaks the whole practice session. How did the first person who noticed tell you about it?
@giangptx A win post gets congrats; a doubt post gets the advice you didn't know to ask for. That trade is underrated. What was the advice he gave about finding the right people?
@alessandro_afl 51 of 394 next to the Fantastical asks is a nice pairing: the requests said what to build, the usage said it wasn't just a loud few. Did the people who asked turn out to be in that 51?
@m0agh When nobody can say what they were trying to do when they hit the gap. A request with a clear job behind it usually holds up; an 'it'd be nice if' tends to fade once you ask. What was the last one you stopped building?
@hi_gaganthakur Signing up and sticking around almost feel like two different products. What did the people who kept using Sunle do early on that the others didn't?
@Bajanpunkw Holding the leaderboard until the numbers come straight from Stripe makes sense. A ranking of self-typed MRR would mostly reward whoever rounds up hardest. What did the person who asked for it want to use it for?
@ArtemYakovlev The report was technically wrong and still pointed at the real problem: the app sounded sure when it was guessing. A ~ is a nice light way to fix that without cluttering the screen. Did he write back after 1.5.2 went out?
@pggalaviz A settings page for the config flag is the part that rings true: each layer is cheap to accept and expensive to own. Typing it all by hand used to say no for you, and that friction is gone. What's one 'future flexibility' add you ended up ripping out?
@vincenzo_naked I'd build the smallest thing that solves what those 100 people are trying to do, which often isn't the feature they named. A hundred asks is strong proof of the problem and much weaker proof of their solution, so the part you hate might be the part you get to skip.
@Plusoneliftsapp Shipping iPhone and Watch together on day one, solo, is a lot of surface. "Tells you how to beat last time" is a sharp hook for lifters. What made you settle on that as the core idea?
@dinocorreia09 The "1 in 100 users asked for it" bit is the real trap: one well-written request can feel like demand. Writing down how many users you expect it to serve also gives you a number to check against after it ships. Has a feature ever come in well under the number you wrote down?
@Jonhrmget Getting a chat app in front of real people is when you find out which parts they actually lean on. Has anything a tester said so far changed what you're working on next?
@IsaakSmart My guess is the hard part is that building always feels like progress and a user call doesn't, so the call keeps getting pushed. Once you started talking to Idearify's users, what was the first thing they said that surprised you?
@HashamBuilds That question works because people answer it from memory instead of imagination. "What would you want?" gets you a wishlist; "walk me through the last time" gets you the workaround they already live with, which is usually the thing worth building.
@DesignByMarso Waiting for the idea everyone would like is such a common trap, and more ideas than time is the better problem to have. With templates, components and tools all competing now, what made the last thing you shipped win?
@IndieBeaverHere Tracking all 18 steps and cutting the ones where people dropped is the unglamorous work that actually moves onboarding. Which step had the biggest drop before you changed it?
@StunLabs Local sync first for me. Owning a copy of my prompts is what would make me trust putting more of them in, and memberships only pay off once creators have a library worth supporting. Which of the two did your first user actually ask for?
@solomonuche The decisions-for-users cost is the one that's easy to miss: it mostly lands on new people who never asked for the feature, while the person who did ask is happy. Has there been something you actually cut from Hintberry?
975 Followers 736 FollowingThird-time venture-backed founder. My second company makes over $100M in revenue and raised $15M+. Now building https://t.co/DB1H3HoNBi in public. I share every detail.
5K Followers 4K FollowingGrow your Business. Higher. Faster. Smarter.
At https://t.co/wzIT2VQklX, we help businesses grow with lead generation, digital marketing, conversions & sales.
4 Followers 60 FollowingTracking AI signals from GitHub, Reddit, Product Hunt & Hacker News.
Finding emerging AI opportunities before they become crowded.
Building in public.
4K Followers 954 FollowingFounder & CEO, MarkGrid AI · Managing Director, Two99 Consortium · Founder, Binary Wall | Building the operating system for modern marketing across the USA, GCC
2K Followers 440 FollowingBuilding things by day, scaling them by night
Founder of GradIntel (AI for grad-school admissions)
prev: @ucla | ex-London banker
1K Followers 514 FollowingBuilding StoreAgent — AI that runs your WooCommerce store from inside WordPress. 1M+ stores use our Rymera plugins (Wholesale Suite, Advanced Coupons +4).
63 Followers 59 FollowingMy Founder Journey in Visual AI and Malleable Software
AI should be more than a chat box. Making the AI Interface of the future.
26 Followers 128 FollowingPM building real apps with AI - no code, no team 🚀
Sharing every win, fail & prompt. #buildinpublic 🌱
YouTube & Instagram: @ipekbuilds
79 Followers 120 FollowingBuilding OpenPumta in public(300+ users)
https://t.co/kdfHgtTdPs
CS Student
Documenting software engineering,
startups, and mistakes
Open to internships
3 Followers 3 FollowingTrack your startup journey in public ✦
Milestones · Revenue · Users · Updates
One link. Your full story.
🔗 https://t.co/XF77RHxZF8
125 Followers 307 FollowingWe are a PC clan for Battlefield and many other games like No Man Sky, Fallout 76, ESO, GTA. Check out https://t.co/RdG1guGTbv for more information.
579 Followers 420 FollowingStaff Product Engineer @blinkist . I also build stuff when my kids are asleep and share my journey about it. a11y. AI. Product
666 Followers 410 FollowingVibe Coding Until I Can Move Out of my Mom's Basement
CoLateral: The All-in-One Engineering Workspace: https://t.co/BrQSTk49CU
976 Followers 391 FollowingSolo dev, 24. Building Stashr from $36 MRR to $10k so I can quit the 9-5. Every bookmark you've ever saved, in one search. Numbers posted daily.
8K Followers 63 Following$550K+ a month across my companies 📈
Run by 80 Claude Code agents (and me, barely)
Happily married w/ 4 kids
https://t.co/OnHkKmIcE6, https://t.co/CFqR7LG3yw, https://t.co/z2mRjzV1gl & more
4K Followers 3K Followinga curious human who loves tech and talks around it 👩🏻💻 • product / marketing
• building; https://t.co/YnZ1HoXNkC
• spreading : https://t.co/ZQ8eBmINKj
5K Followers 499 Following🇫🇷 Founder - building my ideal life x having fun 🎸🎮
💻 https://t.co/kfYZd264dY
🍎 https://t.co/byUyZy5tTa
+ 8 SaaS + 3 mobile apps
317K Followers 8K FollowingFounder and CEO of @acquiredotcom. https://t.co/wRMIssDmhl has helped 1000s of startups get acquired and facilitated $500m+ in closed deals.
84K Followers 1K FollowingLaut BILD “Ex-FDP-Influencerin” // Meinungen sind von mir und so gemeint // sozialliberal // Grün // Referentin im Landtag // hier super privat