Anton

Been sticking my head deep into an SEO this week, still going but a few things are already sticking: (1). pages must map to real user intent, not "AI wrapper" fluff (2). don't create random "AI tool" pages, build around actual workflows people search for (3). one strong ungated demo showing a real product outcome beats 10 thin tooling pages, cold traffic will come but it churns fast if there's no substance (4). if you're exposing AI endpoints for demos, lock them down early, rate limits, captcha, spend caps, learned this one the hard way Still processing this in regard to how I should integrate this into my landing, but we movin, even if its just a bit Anybody else been gone down this rabbit hole? Whats ur approach? #buildinpublic #seo #saas

What actually compounds from building in public folks? (1). credibility, every post is a small proof of work (2). learning, you go deeper when people are watching and questioning (3). relationships, people remember who showed up consistently

@alexcooldev Shadowban on X even though the algo has changed is an excuse to me. Easy to blame the platform when u gotta focus up and polish that content

Lesson 3: A lot of people think that when they use a VPN and get stuck at 100–200 or 800–1000 views, it means they’re shadow-banned. That’s not entirely true. Try posting on your own account in your own country you can still get stuck at 100–200 or 800–1000 views. That’s completely normal, this is content issues. TikTok is just much harder now than before. If you’re using a free VPN or a shared IP with too many users, you’ll likely get banned right when creating the account not shadow-banned later.

@kzitouni1 500 in 3 weeks is insane. Consistency did that! Was it just slow and steady or did u have some post moment?

I started posting on X 3 weeks ago - 0 likes - 0 friends - stalled account Tonight we crossed 500 follows I just wanna thank the community here I'm following a few new faces too, what are you building?

@Dwriteway Yea waiting to be ready is just going in circles. You only learn by doing and doing and that phase u just suck is where u really really learn

“Being ready” is a myth. You start. You suck. You figure it out. You get better.

@sharran Yea lots of people polish theyre work environment and completly ignore the information environment. Normalization needs to happen on that domain too

You are designing your environment. Your environment is designing your future. The content you consume, the people you engage with, and the conversations you entertain shape what feels normal and possible. Normalize higher standards.

@stijnnoorman Yea its also about that curiosity is something which becomes consistent, and also not just consuming new stuff but sort of go in with a testing mindset where u actually assume ur wrong so u can update consciously

Dumb people stop learning after school. Smart people keep learning their entire lives.

@alexcooldev And lots of builders have conviced themselves too that this vitamin is the actual pain killer here. Takes authenticity to admit that

Unpopular opinion: Building an app with just one feature that solves a real pain point is more likely to succeed than an app packed with too many features that are just “vitamins.”

@Layton_Gott Haha, guilty. The idea list keeps growing into a larger and larger graveyard of possible apps.

Hot take: 99% of SaaS startups DON'T fail because the idea was bad. They FAIL because the dev got bored and started a new project.

@theo_jil Yea, niche audience will grow ur product further better too. just a few right people who are really locked in on ur idea + word of mouth > 1k lukewarm who drop after a week

You don’t need a big audience. You need the right audience.

@trikcode Dont all the max people share infrastructure too? So it depends what all the other max people to at the same time and steal each other bandwidth and that makes them even pay more? 🤣

Apparently $200 isn't it enough People paying $200/month for Claude Max are burning through rate limits in 2 hours. $200/month. 2 hours. This is not a subscription. This is crazy!

@IAmAaronWill Feels ike it doesnt it? Everybody just scraping the same top posts and it just becomes one big slump. Slowly bleeding out originality

Old way of creating content: 1. Figure out audience's problems 2. Come up with solutions 3. Write 10 tweets 4. Hate them all 5. Rewrite them New way of creating content: 1. Have a system scrape last week's best posts 2. Have it generate 10 ideas 3. Pick the best and edit

@vishaldeshpande And smart peope use it to build elaborate and wrong ideas confidently haha

the dumbest person you know is being told "you are absolutely right" by an AI right now!

@housecor Yea its all pattern based and if youre agreeable with the LLM to it wont change that. Think personality mightve now moved to the prompts 🤣

An odd thing I'm noticing in 2026 - With AI, everyone's PRs feel more similar. It used to be easy to tell who built what. "Oh, this is John's quirky code". Now with AI, the codebase "feels" like it's written by *one* person. More consistency. Less weirdness. Fewer quirks.

@Tofunmithedev Private repo era going hard here. We know the step though: SHIP. THE. UGLY. PRIVATE. REPO.

Ignoring everyone's call because I'm building a million dollar b2b saas that will sit in a private repo on github making exactly $0.

@tomfgoodwin Yea agree cold email is really really hard to surface. And AI spam aint innocent on that. But warm email probably still works if you've built the distribution and audience first

Am I right in thinking that AI completely destroyed email marketing? I don't think anyone actually reads any emails anymore. Other than from existing known people on existing work.

@sethbradleyesq @russellbrunson Thats a better framing actually

@antonships @russellbrunson Most people don’t fail, they just never start because it “looks dumb.”

The ideas that seem crazy at first are often the ones worth exploring.

@katarinaore I would. churn happens before users even get to the part that would've made them stay, the app being good is kind of irrelevant if onboarding doesn't get them there

Working on onboarding for my app People say it's more important than the app itself.. Would you agree?

@LoicBerthelot accurate, except sometimes you wake up mid-implementation and they've already dropped something else hahaha

life as an entrepreneur: > wake up and see what anthropic launched > implement > go to sleep > wake up and see what anthropic launched > implement > go to sleep Rinse & repeat 7x/wk until generational wealth

@svpino Ok verification seems not like an endpoint you maybe arrive at but it’s something u have to consider from the get go. So that and ALSO evaluation what comes as answer after the retrieval. That is really something uve got to consider once u start dabbling with knowledge based.

RAG is much more than a search engine. This is one of the most common misconceptions I have to help people get out of their heads. Think of RAG as a "retrieval + reasoning pipeline" that requires constant evaluation for it to work. Retrieval without evaluation is just fancy ctrl+F. Your agents need to judge what they retrieve, not just fetch the top-k nearest vectors. Here is the TLDR; every RAG system must have a verification layer.