James

@delveroin Building signals.microbuilder.co Free tool for indie hackers and vibe coders to find opportunities to build. Scanned from 9 sources and updated daily

Its Friday Founders share what you are building

@sethsetse went through this last year. 14-day review, rejected for a screenshot being 2 pixels off. launched a web version same day it got rejected. web version made money within the week.

“sir they’ve paid $100 for the Apple Developer account” “good. wait 2 days to confirm their account then reject the app 10 times over the next 20 days. if they make it through take a 30% cut of the revenue and pay them out 2 months later. ban them if they try using Stripe”

@DanKulkov bookmarked. spent 6 months convinced i was bad at marketing. turned out i was just bad at the marketing i was trying. switched to one channel, one format. completely different results.

How to do marketing If you are a Solopreneur who sucks at marketing docs.google.com/document/d/e/2… Free Google Document

this is insane > type > "5 unexpected high-fiber foods you should eat" > wait > 40-sec video scheduled new saas soon this is insane

The scoring: Strong Demand (people complaining AND paying for bad alternatives) + Wide Open (no funded startup owns it) + Weekend Project (solo dev can ship an MVP in days). I built this to find my next product. Scanning manually was taking hours a day. Made it public and kept it free. Check it out: signals.microbuilder.co Updated daily. No signup required.

@steipete @badlogicgames The 'drives all the claws' framing is right. Agentic workflows finally have an underlying runtime built for them, not bolted on. I built the same orchestration logic 3 different ways in 18 months. Each time the tooling matured and made the previous version obsolete.

At AI Engineer London, watching legend @badlogicgames talking about pi, the component that drives all the claws! pi.dev

Same pattern. I hit the Claude Max cap 4 times this week. The issue is not the context window, it is running multiple parallel tasks. One agent writing code, one reviewing, one scanning for bugs simultaneously. 00/m starts feeling tight fast when you are actually using it for real work.

Many people asked Why. I have multiple friends telling me to move, most of them are heave CC users before. No other friends ask Codex people to move to CC. But that is not enough tho. Recently, I started hitting my Claude Max ($200) limit quite often despite results are not getting significantly better. Also multiple tools like OpenCode, OpenCode, etc., don't have "official" support for Claude OAuth anymore, whereas Codex fully supports these tools. I guess it's time to at least give it a try. 😄

Switching to Codex next month when my Claude Max sub expires, it’s finally time.

@andrewchen Builder perspective: the 50 tok/s frustration disappears when you stop waiting and start parallel tasking. I run 3-4 Claude sessions now. One writes code, one reviews it, one drafts tests. Total throughput is not that different from a faster single stream.

People that talk at 10 tokens/s = very smart LLMs that output at 50 tokens/s = annoying and slow 🤷‍♂️

@matt_gray_ The other 90%: agent workflows that run while you sleep. I have one that monitors competitor signals, another that scans for demand patterns, another that drafts content. All built in 2-3 hours each, run 24/7. That Ferrari works night shifts now.

@imbhaveshkumarr @marclou zero coding experience is actually an advantage early on. you're not overthinking architecture, you're just solving the problem

@Microbuilderco @marclou mann this is crazy way to look at it, this is my first ever saas with zero coding experience lol but def gonna implement this

didn’t expect getting an offer on TrustMMR in only 2 days all thanks to @marclou idearupt.ai has surpassed 600+ users in 3 weeks, and so far 4 paid customers. this is gonna be huge if you were me would you sell it?

@DanKulkov complaint threads on reddit > any ad spend i ever tried

How to do marketing If you are a Solopreneur who sucks at marketing docs.google.com/document/d/e/2… Free Google Document

no favicion in 1 month what god do i need to worship to finally see it? >danrecommends. com/favicon.ico >48px

@tibo_maker the bit about feeling like 'the lucky guy who got acquired once' is the fear nobody admits but every founder carries. the $1M/month is great. the fact you shipped 5 things before you got there is the actual story.

yes, I crossed $1,000,000 / month was trying to keep it quiet but it got leaked on a pod so... let's go quick backstory: I sold Tweet Hunter and Taplio for $8m in 2024 then got back to work and honestly, I was terrified when you set a standard for yourself, you have to live by it and that pressure is real every single thing I shipped felt scary - I felt like a fraud, "the lucky guy who got acquired once" every launch, I kept thinking: what if it flops? what if I go from "the guy who exited" to just... a nobody? so I forced myself to ship a ton a lot failed I was embarrassed by the bugs in my MVPs but I kept going Revid, Outrank, SuperX, PostSyncer, Feather -> five products thriving today they do more than anything I've ever built before grateful to my co-makers, to the team, and to everyone following along, giving feedback, using the products every day thank you, genuinely 🙏

@rcmisk @yuuukta building in public has moved the needle most. not just shipping updates but showing the process, the failures, the weird edge cases. people follow because they're doing the same thing and want to know they're not alone.

@yuuukta what are some distribution tactics, or strategies you are using!?

real talk: why haven't you shipped yet? A) still adding features B) not sure it's good enough C) no idea how to get users after D) already shipped, stuck on distribution drop a letter

@hamptonism nah you don't need to clone the whole terminal. pick the one feature that makes it worth $30K for a specific vertical and charge $300/month for just that. legal professionals alone would pay it.

Founder of the Bloomberg Terminal explaining the function that makes it worth $30,000 a year:

@cgtwts this isn't really a vibe coding problem. it's deploying AI-generated changes without the same review gates you'd require for human code. different cause, same gap.

Silicon Valley was really so ahead of their time

@GeoffreyHuntley spent time this week testing whether my content shows up in ChatGPT/Perplexity answers for target queries. it doesn't. there's no tool that tells you why or what to change. that gap is massive.

AI SEO is now a thing

@agazdecki been here. built 4 in a row before it clicked. the bottleneck was never the code, it was whether anyone actually needed it before i started.

How founders feel when vibe coding their 15th SaaS with zero users:

@josefbuettgen tried something similar for a niche vertical last year. interface took 3 days. getting the data pipelines to not fall over took 3 weeks. bloomberg's moat isn't the terminal.

the urge to build this in a weekend with claude code + opus 4.6

@DanKulkov looking forward to it. curious what's doing the actual work on launch day, PH or your own list?

new app we are cooking #1 soon

@TheGeorgePu honestly the BYOD twist is the more interesting product than the managed version. selling a software subscription on hardware you don't own is harder to justify. i'd just run the stack myself at that point.

Perplexity just launched their 'Personal Computer'. It's a Mac mini they bought that runs for you 24/7. Your files, apps, sessions - routed through them. But a Mac mini costs just $599. Why pay Perplexity monthly for a computer you could just own? The product isn't the computer. You are.

@rcmisk D, always D. built 3 tools last year with zero distribution plan. the product being done felt like the finish line. turns out it was the starting gun.