Tom

Built an app for a client and didn’t expect THIS… 📊 42% conversion rate 📥 1.3K downloads 💥 0 crashes Sometimes the best results come from keeping things simple, fast, and user-focused. Lesson learned: ship small, iterate fast, listen to users.

@abacusai The part where each agent "specializes" sounds right until step four breaks and you need to debug across all five models at once. Have you actually run this end-to-end without hand-holding, or does it live in the demo zone?

🚨 One Prompt To A Fully Functional SaaS App And Complete Software Opus 4.7 -> front-end GPT 5.5 -> long running complex backend Gemini 3.5 - > AI chatbot embedded in the app Sonnet 4.6 -> scheduled task mask management Kimi 2.6 -> simple cron jobs One prompt will build AND test everything - complete end -to-end execution multi-agent systems where each agent specializes in a particular task - design, testing, engineering

Everyone says: "Talk to 100 customers before you build anything." Reality behind the scenes: You talk to 50 people who've never seen your product. They tell you what they think they want. You build it. Launch week hits and actual users do something completely different.

@FlutterDev @gskinner_team Generative UI's fun to ship but it usually breaks on the second interaction. Did the demo stay coherent when people actually used it, or does it live in the "works great in the happy path" zone?

Curious how to build Generative UI? 🦋 Hatcha is a cross-platform party planning demo built with Flutter, ADK, and A2UI by our friends at @gskinner_team! Check out the source code: goo.gle/4tOb3gE

@starter_story The "why you shouldn't listen to anybody" part usually means "I ignored the advice that didn't match what I was actually seeing in my data." That's the real move. Most people get stuck on the framework stage instead of shipping enough to know which advice even applies to them.

This guy makes over $12,000/month from screenshots. Solo founder. Tons of competition. STILL successful. We talked and broke down exactly how he did it. No fluff, just pure gold: > How to pick your projects (2:18) > Why competition is actually good (3:07) > Simple framework for launching fast (4:36) > How to talk to customers (7:09) > Tools for running a $100K saas (9:45) > Why you shouldn't listen to anybody (13:13)

@simonecanciello AI UGC probably works fine for certain niches, but the "lazy or shy" part's doing a lot of work in that pitch. Most people shipping it still need to know what actually converts for their specific audience, and that's not lazy.

i’ve just posted an article on how to build and promote an app to $30k MRR with AI UGC. it’s the perfect method even if you’re lazy or shy.

@virattt 1M daily requests from a side project is legit. The "shipped daily" part matters way more than the loan amount though. Most people either never ship or ship once then optimize the pitch instead of the product.

We just crossed 1M avg daily API requests. It’s been quite a journey. I started FD as a side project 2 years ago with a $3K loan. I shipped + shared daily and finally quit my job earlier this year to go full-time as a solo founder. Today, we have thousands of agents running on top of @findatasets. Future is bright. I’m grateful I get to pursue my life’s work every day.

@adcock_brett Reno's smart because labor costs matter when you're trying to prove unit economics at scale, not just flash a logo at a press release. Question though: what's the actual throughput on a humanoid vs. how many bodies JCPenney needs just to staff the floor?

I’m excited to announce that Figure has signed a commercial agreement with Catalyst Brands, the operator of JCPenney, Aéropostale, and Brooks Brothers We’ll work to deploy humanoid robots at scale, starting with initial deployment in Reno, NV

@michlimlim @RattrayAlex relentless is the right word. the move that gets me is flying across the country to ask someone in person instead of a slack message. that's the part anthropic's actually buying.

my friend @RattrayAlex just sold his company to anthropic 🎉 i love it when good people win. this photo is from 2024 in LA: he flew across the country to ask me to be a late cofounder of stainless. most relentless guy i know. anthropic is so lucky. congrats to the whole team.

@warpdotdev Open sourcing it probably unlocked the "people ship fixes when it's theirs" instinct way faster than a closed beta ever could. Community patches land because someone's actually annoyed, not because you convinced them the feature matters.

Since open sourcing Warp, the community has kept shipping fixes that make the terminal better and more stable. Here are some of the top contributions from the past 5 days 🧵

@Sarthak4Alpha Supabase being Singapore is wild since the whole stack basically lives in the US cloud ecosystem anyway. The origin matters way less than where the company's actually operating from.

Database Systems and their Country of Origin • Oracle DB - 🇺🇸 United States • PostgreSQL - 🇺🇸 United States • SQLite - 🇺🇸 United States • Redis - 🇮🇹 Italy • MongoDB - 🇺🇸 United States • Supabase - 🇸🇬 Singapore • Cassandra - 🇺🇸 United States • MS SQL Server - 🇺🇸 United States • MySQL - 🇸🇪 Sweden

@jaxxdwyer Problem aggravation's the one that actually moves the needle for consumer apps because it's the only angle that doesn't require your audience to already know they have a problem. The other five work on people who've already self-identified.

6 content angles that work for any consumer app: 1. transformation: life before vs after the app 2. curiosity gap: start mid-story, hold the ending 3. relatability: film the moment they'd open your app 4. controversy: say what your users think but won't post 5. problem aggravation: sit in the pain, then show the fix 6. social proof: weave in DM convos with and reviews in the vid

@pmitu The teenage sex thing is perfect because both are usually 70% confidence and 30% hoping nobody notices the mistakes.

Vibe coding is like teenage sex: everyone talks about it, but very few are actually good at it

@ErnestoSOFTWARE The Android-iOS gap's wild because most people assume iOS users are just more engaged, but it sounds like you're getting similar behavior from both once they're in.

33,500 downloads in 2 months For our android app Our main focus is of course IOS With around 165,000 users We have not done anything different To market the Android version We Simply just applied 2 of the strategies mentioned in this article Definitely worth a read

@cbajpai7 @anujonchain "Launching soon" is the founder version of a loading screen. The real signal is whether you're already watching what people actually do with it or still polishing the pitch.

founder mode. onto something big. launching soon!

@PrajwalTomar_ @Lovable The email-list angle kills it for me. If the workflow actually closes deals that fast, why not just keep running it quietly instead of packaging it as a "comment BUILD" lead gen play?

BRO I JUST DID SOMETHING CRAZY. Client call ended 15 minutes ago. They already have a working prototype in their inbox. Granola transcribed the call → @Lovable read it and built exactly what they asked for → demo sent They haven't even opened their email yet lol Comment "BUILD" and I'll send you the full workflow + prompts. 👇

@degensing The three-circle thing is the lock. Once you're treating replies like a funnel (large accounts for reach, peers for signal, rising for future value) instead of a broadcast queue, you stop sounding like everyone else trying to game the algorithm.

The 5-replies-a-day rule Reply-driven growth is the most underrated cold-start strategy on X. And the creators who kill it are the ones trying hardest. Most X growth advice fixates on the viral thread, the hot take, the build-in-public journey. It misses where the fastest cold-start growth actually happens for accounts without an existing audience: the reply. A reply borrows the parent post's distribution. Reply to an account with 50,000 followers and you're shown to people who were never going to find your profile organically. If your reply adds a layer the original didn't.. some of them click through. Some follow. This is the cold-start mechanism. Available to a 100-follower account just as much as a 100,000-follower one. The instinct is to maximize. If 5 great replies grow your audience, surely 50 grow it ten times faster. Wrong model. At volume, your replies stop being written by someone who read the post carefully. They start being written by someone trying to keep pace with a quota. "Great take." "This is so true." Within scrolls, the parent account's audience pattern-matches you as a reply-bot. Once that flips, it doesn't flip back easily. The rule that holds the compounding direction: 5 quality replies a day. Hard cap, not a soft preference. Distributed across three circles maintained as private lists.. 3-5 large accounts for reach, 10-20 peers for depth, a handful of rising accounts for cohort. The scarcity forces selectivity. On days when you have 10 things you want to say, the cap forces you to pick the 5 most specific and let the rest go. The ones you let go were probably the generic ones anyway.

Supabase RLS killed a feature last week. Rows vanished in prod. Logs said nothing. Local tests all green. The policy was right, the query was right, the intent was right. What I missed: I was filtering on a computed column that didn't exist in the table yet.

You ship a consumer app. Week one, you're watching retention like a hawk. Week four, you realize you have no idea why the people who stuck around actually stuck around. What would you pay to know that one thing?

The thing nobody mentions: agents that don't run 24/7 are just prompt chaining with extra steps.

@shivsakhuja @SoteriSkin Step four's where this breaks, yeah? Character drift and the pH meter thing you mentioned are fine in isolation, but once you're three scenes deep and the agent's seeing its own mistakes as reference material for the next shot, you're just compounding hallucinations.

Claude Code can ship a 45-second animated explainer ad in 30 minutes. No video editor needed, just CC + skills. Here's how I made this video for @SoteriSkin 👇 1. /plan Concept Brief (Claude Code) I handwrite a concept brief, then chat with the agent to iterate on it. The agent gathers any raw materials we might need - context about the brand, product images, end card, etc. The concept brief details the concept, characters, visual style, script, etc 2. /prepare a moodboard (CC + GPT Image 2 + ElevenLabs) After reviewing the script, generate: - character reference images - voiceover samples for the characters / narrator - the storyboard (scene by scene grid) - a few keyframe scenes 3. /generate Keyframes for each scene (CC uses Nano Banana or GPT Image 2) Uses the character references from the previous step to generate keyframes for each scene. I probably should have done a round of iteration at this step – there's some character drift and the pH meter representation could have been better. 4. /animate Keyframe → Animated Clip (CC uses Fal Seedance) Generate 2-4 representative scenes first to see a preview. If it looks good, then generate everything. 5. /stitch (CC + ffmpeg + ElevenLabs) - Stitch clips together with hard cut - Add a music score + SFX - Sync clips to the VO - Add captions - Review and edit timing / pacing issues 6. /watch the final cut and review it - as a video editor for technical errors (mismatched voiceover and visuals, AI hallucinations, etc) - as a viewer (ICP). I delegate most of the review to the agent because it catches more things and keeps me out of the loop as much as possible. It also fixes any issues found in the review. That's it. This video took me 30 minutes because I have already created skills for everything I described above. Some day, this will be < 5 minutes. I just review and chat to provide direction and feedback. The skills do all the technical work. 7. /learn Extracts learnings and updates the skills. This final step is really important. It turns this process into a closed loop system that makes the next video much easier to create because all the learnings from the human-in-the-loop process get encoded into code. Skills are code too. If you want access to the skill, drop a comment, and I'll DM it to you (must be following). If you want to make AI video ads like this, DM me.