Finding signal on Twitter is more difficult than it used to be. We curate the best tweets on topics like AI, startups, and product development every weekday so you can focus on what matters.
Originally published by @EXM7777 on X. Tech Twitter preserves the original source alongside this readable edition.
Before you dive in
• Learn the 12-month blueprint to grow from zero to 100K followers on X: capture daily value, master your niche, stay consistent, build real relationships,...
Best for builders who want practical takeaways. 21 min read.
This is the blueprint I followed like a retard for a full year... it helped me build a $50K MRR side business, the strongest audience in the AI niche, and some fucking cool friends.
Here's what you're walking away with after this read:
a 5 step daily content loop you can run today with a notebook and a draft folder, no apps, no agents, no api keys
a 4 criteria filter for whether a tweet is worth shipping, the same filter we now use to grade million dollar long-form articles
the value tax, the math behind why the first 6 months of public writing are non negotiable, and what changes inside your head once you survive it
the boundary line for ai in your content workflow, the places it's allowed to touch, the 3 places it never gets to touch, and the popular shortcut that turns your account into slop on a 90 day delay
a no-bs read on what 365 days of public writing actually cost me, and what it didn't cost me that everyone assumes it will
if you came here for “the 5 ai prompts that grew my account 100x,” this is not that piece, this is the boring version that actually worked
i'll actually share an agentic content creation system for X shortly in my community (weeklyaiops.com), join us if you're into that
you might think i'm an engineer, i'm not
a lot of people who follow me assume i'm a developer, a crack at technical stuff, someone who came up through code
i'm not, i come from sales
a lot of people also assume i'm a native english speaker, because the writing converts and the cadence reads clean... well, i'm not
if you're reading this thinking easy for him, he's a developer who can build his own tools, easy for him, he writes in his first language
neither is true, the unfair advantage you're projecting onto this account doesn't exist
what i had on day 1 was the willingness to lock in on this since the day chatgpt became actually usable, that's it
people seek exotic explanations because mundane ones implicate them, if the answer is i locked in for 12 months, every reader knows they could have done that and didn't, the engineer myth and the native-speaker myth let them off the hook, this article doesn't
We doomscroll, you upskill
Get the 10 tweets shaping how builders think today.
Newsletter
We doomscroll, you upskill
Get the 10 tweets shaping how builders think today.
studying tweet anatomy is the trap, hook formulas are the trap, optimal posting times are the trap, all of it is real and all of it is downstream of something else
the something else is what you do with the other 23 hours of the day
if you want to grow a strong audience, the kind that buys from you, attracts brands willing to pay premium for your content, builds a network of operators you actually want in your dms... you have a lot of work to do outside the platform
run a real business, go super deep into your niche, do the things others are too lazy to do, talk to people, live interesting things and capture them as they happen, problems your customers are facing, how you fixed them, what you learned shipping the thing nobody asked for
the platform rewards people who show up with real material to translate
the platform punishes people who show up empty and try to manufacture content from nothing
the rest of this article is the system that turns the off-platform work into on-platform results, 3 outside-x pillars, 4 inside-x disciplines, and an ai layer that comes last, not first
pillar 1, capture daily value from real operations
my number one piece of advice for growing a brand on x is to do things outside of x, then capture them obsessively
i started with a pen and a notebook, walking around all day, scribbling raw thoughts, half sentences, problems i ran into, solutions i tried, things customers said that surprised me, ideas that hit me in the shower
at the end of the day i'd sit at my desk and turn the rawest of those notes into ideas, and the ideas into polished tweets in typefully
that was the whole content engine for the first 6 to 8 months, no apps, no agents, no automation, just paper and a desk
then the stack evolved
now i capture moments with voice notes in telegram, straight to my hermes agent, which syncs each thought into a structured knowledge base inside obsidian
i don't ever read obsidian, i have claude code pull the best ideas every morning and i sit with it on how to articulate each one to make it worth a million dollar
notice what changed and what didn't, the tools changed, the principle is the same as the notebook days, capture daily, polish at the desk, ship the next morning
but once you commit to capturing your life this way, the way you live starts to optimize for things worth capturing, you start picking the harder customer call over the easier one, the new tool over the comfortable one, the conversation that scares you over the one that doesn't, the act of capture quietly rewires the act of living, and that's the actual moat, not the obsidian vault
if you're starting today, you don't need the agent stack to begin
the agent stack only earns its keep after you've proven you can fill the bucket every single day with raw value, the bucket has to fill before any pipe is worth building
pillar 2, lock in on your niche, ship novelty before anyone else
if you're writing about ai, you cannot be late, ever
instagram and tiktok have time, you can talk about a workflow from a month ago and it still hits, the algorithms are shape-fitted for evergreen, the audiences are forgiving
x is not that
on x, novelty is the currency, you have to show up as someone in the trenches who saw this coming before anyone else, who built the workflow on the morning the model dropped, who has the screenshot from yesterday not from last quarter
the way you stay in front of that without burning out is by building systems to feed you the edge
aligned news for ai signal, the /last30days skill for what's actually being talked about across reddit / x / hn / youtube / tiktok in the last month, di.gg/ai for sourced briefings on the model and tooling layer, you can equip your agents to pull from these every morning and surface the 3 to 5 things you should be paying attention to today
novelty doesn't come from being smarter than everyone
novelty comes from being earlier, and being earlier is a system, not a talent
the trap inside that system is that the system itself has a half-life, the tools you use today to be 48 hours ahead of the timeline will be commodified in 6 months and everyone will be using them, the actual edge is the discipline of replacing the system every quarter before the rest of the platform catches up, niche-mastery is a treadmill not a destination, and the people who treat it like a destination are the same people who get lapped quietly by the people who never stopped looking for the next ingestion source
pillar 3, read and write daily, off x, to improve the craft
read every day, write every day, off the platform, in long form
books, essays, threads from people you respect, your own private notes, doesn't matter what
what matters is that you're moving prose around with your hands every single day in a context where no one is watching and no metric is being recorded
great writers get noticed on x
bad writers fall into the masses
every shortcut that tries to skip the writing-as-craft step ends in the same place, your account looks fine for 90 days then everyone realizes you sound like every other ai-generated thought-leader account, and you're invisible again
the off-x writing isn't training for the on-x writing, it's the only place where you can write badly without consequence, and writing badly without consequence is the only way you ever learn to write well, every metric on x punishes the bad version of the next sentence you'd otherwise need to write to find the good one, kill the metric-free space and you kill the next 6 months of your craft growth without ever realizing it
practice the craft in private, perform the craft in public, the public part is downstream
the value tax
now here is what you have to do on x
for the first 6 months, you have to pay the value tax
i posted 7 high value pieces per day, every day, for 6 months straight, no exceptions, no off days, no “took the weekend”
that is roughly 1,260 pieces of pure value over 180 days, and yes that number is uncomfortable on purpose
the combination of high volume and ultra high value just can't miss
high volume alone is noise
high value alone is invisible because nobody knows you exist yet
the only thing that breaks through is both at once, executed without flinching, for half a year
now i'm bigger and i can allow myself jokes, some ragebait, some engagement farming, none of which would have worked on day 30 because nobody had any reason to give me the benefit of the doubt yet
i earned that benefit by paying the value tax for 6 months, and you can earn it the same way, the math is the same for everyone
what counts as high value, here is the test i used and still use today:
the post clearly explains how to achieve something specific, “how to build a copywriting agent using x,”“how to master context engineering,”“how to use claude code for y”
anyone in the audience can understand it, no jargon walls, no inside-baseball
it's structured step by step, replicable, not just inspirational
it's novel, the vast majority of readers had no idea you could do this
the meta test, would someone bookmark this and come back to implement it later
if a draft fails any of those four, it gets reworked or killed before it ships
the red eye in the feed, 365 days no miss
i didn't miss a single day for a full year
365 days, 0 misses, same time every day, the big red eye in the feed showing up at the exact hour people had started expecting it
this is the most “obvious” advice in the entire article and it's the one most accounts violate inside the first 30 days, that's the whole game
the audience's brain quietly anchors on a rhythm, when you show up at the same time every day they stop scrolling past the avatar because it becomes a known signal, when you skip 3 days the rhythm breaks and the next post lands with no context
what you're doing is pavlovian conditioning at scale, you're training thousands of strangers to look for a specific avatar at a specific hour, and the moment you break the pattern you have to retrain them all from zero, the cost of inconsistency isn't one missed post, it's the dilution of every post that comes after
consistency is not a personality trait, it's a forcing function on the audience's attention, and the forcing function only works if you don't blink
if your version of consistency includes thursday-skips and weekend-offs, you're playing a different game with a different curve, both are fine, just don't pretend the curves are comparable
two friends, not two hundred
when i started, i had 2 good friends growing in parallel, same mindset, same ai obsession, same trenches
we engaged organically with each other every single day, dropped on each other's posts, qt'd each other's bangers, sent each other early drafts in the dms before they shipped, cheered when one of us cracked something
3 people total, that was the entire founding network, no slack of 200 creators, no engagement pod with 50 randos, no “give a like get a like” arrangement
what that did was simple, when i shipped a long thread or an article-tier post, those 2 friends made sure it got the first 50 to 200 views from real humans inside the first hour, which is enough signal for the algorithm to start showing it to a wider circle
their network organically expanded into mine, my network organically expanded into theirs, no one asked for anything, everyone gave constantly, and within 6 months that 3-person micronetwork had grown into a real distribution layer
the deeper reason 2 friends beats 200 followers in the same window of growth is signal density, 2 friends will read every single one of your posts and tell you exactly what's bad, 200 followers will read 1 in 30 of your posts and tell you what flatters them, brutal feedback compounds harder than warm reach when you're still figuring out what your account is supposed to be
if you're starting today, find 2 people, not 200, who are growing alongside you in the same niche with the same energy, just be friends, the rest follows
give a lot, expect nothing in return
every relationship i built in the last 12 months that mattered started the same way
i gave first, repeatedly, with no expectation of anything coming back
a thoughtful comment on someone's post, a useful intro between two people who didn't know each other, a free hour of advice in a dm, a bug fix on a tool i used, a quote tweet that genuinely helped an account smaller than mine, a screenshot of someone's product showing it working in the wild
the math is brutal, you have to give 50 to 100 times before anyone notices, and the people who quit at give number 4 spend the next year wondering why nothing closed
the people who keep giving end up with a network that, unprompted, refers them to talented operators, introduces them to brands that pay premium, brings deals into their dms instead of forcing them to chase
building relationships pays off harder on x than on any other platform i've used, the surface area for warm reach is enormous, you can dm almost anyone, you can quote anyone, you can show up in anyone's replies and add real value, the platform is unusually generous if your default mode is to give
the structural reason is that x is a high-volume timeline of strangers asking each other for things, attention, follows, sales, sponsorships, intros, and the rare account that reverses the polarity becomes a default-trust node in a network that has almost no trust by default, the giver isn't competing with other givers, the giver is competing with thousands of takers, the math works in your favor by absence
i barely had an ai system for 8 months
now we get to the part everyone wants to hear, and it's going to be shorter and less impressive than you think
for the first 8 to 9 months of this run, i barely had any ai system in my workflow
the bulk of the growth happened with a notebook, typefully, and a brain
if you're reading this thinking the unfair advantage was a gpt wrapper or a custom agent stack, it wasn't, the unfair advantage was the discipline outlined in the previous 7 sections, the agent layer was something i added on top once the discipline was already producing results
the order is non negotiable
ai amplifies whatever you already have
if you have discipline ai amplifies discipline
if you have slop ai amplifies slop, faster
accounts trying to grow with ai today are skipping the discipline step, the result is the same in every case, slop produced at scale, filtered out by the algorithm within 60 days, and the operator wondering why the magic didn't happen
if you only take 1 thing from this article, take this, build the manual workflow first, then layer ai onto it, never the other way around
the manual content loop, 5 steps, do this for 6 months before any agent
here is the entire content engine i ran for the first 8 months, exactly as i ran it
step 1, live your day, and i mean live it, do real work, ship things, talk to customers, fix problems, run your business
step 2, capture what's interesting, what you learned, what would be valuable to someone else, in whatever form you can (notebook, voice memo, slack draft channel, doesn't matter)
step 3, sit at your desk for 30 minutes, expand the raw captures into draft tweets in typefully, no editing for tone yet, just get the ideas out of your head and into the queue
step 4, by end of the day, go back to those drafts after giving them some time, the time-distance is what catches the bad ones
step 5, polish what survives into shippable tweets, schedule the next morning's drop
5 steps, 1 hour total per day, no agents, no api keys, no obsidian vault, just you and a draft folder and the discipline to not skip the time-distance step
if you can't run this loop for 30 days straight as a manual workflow, you're not ready for the agent layer, the agent layer doesn't fix discipline gaps, it amplifies them
where ai is allowed in the loop
once the manual loop is humming for 6 months, ai earns a seat at the table, and the seat is narrow
these are the places ai is allowed to touch my content workflow today:
articulation, when an idea is right but the words aren't hitting, ai is good at restructuring the same idea 3 different ways so i can pick the one that lands
accessibility, when i'm writing about something dense, ai is good at flagging where a non-expert reader would fall off, and rewriting that section in plainer language
vocabulary variation and typo passes, low-skill, high-frequency, ai is fine here
post-implementation research, after i've built or shipped something, ai helps me capture the relevant news, context, and sources and weave them into the post, the after is the whole rule
knowledge base maintenance, claude code keeps my obsidian vault structured and linked so i can pull on it daily without ever opening obsidian myself
recycling, claude code surfaces old captures from 6 months ago that are suddenly relevant again because the model landscape changed, and i rework them into new posts
long form structure, for articles like this one, ai helps me outline, source, and pressure-test the spine before i write, the writing itself is still mine
testing speed, ai helps me draft 3 versions of a hook or a thread cover so i can ship the one that wins, faster
every one of those has the same shape, ai sits downstream of a real captured idea i already have, ai doesn't originate the idea, ai helps the idea hit harder
where ai is never allowed
3 places ai never touches my content, each one non negotiable:
writing tweets from scratch using my voice dna, this is the most popular use case being sold right now and the single fastest way to ruin an account, you'll save 30 minutes a day for a month and watch your engagement collapse, every post will sound vaguely like you but stripped of the small surprises that make it actually you, the audience can't articulate what changed but the bookmarks dry up and the dms go quiet, you'll end up posting slop on a 90 day delay
originating news, the second your readers learn you're translating headlines for a living instead of shipping things and reporting from the trenches, you've lost the only edge you had, the implementation has to come from you, the news layer is just the wrapper
the live-and-capture phase, ai cannot live your life, it cannot have the customer call, it cannot be in the room when the model breaks, it cannot ship the workflow that taught you the lesson, that part is yours forever, and that part is where 90% of the actual content fuel comes from
the failure mode in all 3 cases is the same, ai produces output that looks correct from the outside and is hollow from the inside, the audience won't catch it for 30 days, then they'll catch all of it at once, and the trust takes 6 months to rebuild if you ever rebuild it at all
what 365 days of this actually cost me
most creator-cost essays dramatize the toll, half of them sound like therapy in public
mine is going to be unusually short
the only real cost was that i'm more chronically online now
when i started, i was spending 30 minutes per day on x and 30 minutes per day writing content, 1 hour total
today i scroll closer to 1 hour to 1 hour 30 daily, plus the writing, the scroll-time has roughly doubled
that's the entire cost ledger, no relational damage, no mental health spiral, no identity-pressure crisis, no financial cost, no friend-group rupture
the typical creator costs didn't show up because the bet was intrinsic from day 1, the work was rooted in real business operations not in a manufactured persona, and the volume of value going out (7 a day) outweighed the metric anxiety coming in
i never almost quit
i know this is the part of the retrospective where the writer is supposed to insert the dark night of the soul, the week they almost stopped, the dm from a hater that almost made them log off forever
i didn't have one
not a week, not a day, not a moment
so my honest answer to “how did you stay locked in for a full year” is unspectacular, you just have to work with the right people, give it everything you have, and keep pushing
if you're looking for a more romantic version of that answer, i don't have one
day 365, the new fear
i'm not the same person who started this run
on day 1 i was afraid i couldn't articulate my thinking in short form, confident i had the underlying ideas, betting that if i could solve the medium the ideas would carry me
today i can articulate it, i think i know how this thing works better than 99% of people on the platform, and i'm playing a different game, the underdog game is over, the leader game is starting
the new fear is that the platform itself is changing under us
in 12 months, 80% of our feeds on x are going to be bots, the signal-to-noise ratio is going to collapse, and the question shifts from “how do i grow from 0” to “how do i stay differentiated when half the timeline is synthetic”
when 80% of feeds are bots, the value of writing well stops being about quality, the bots can write fine, the value becomes verifiable presence, the proof that there's a real person behind the account who shipped something this week, took the customer call, has a face and a voice and a track record across other channels you can cross-check, the framework in this article isn't about content output anymore, it's about being unfakeable
i don't have the full answer yet, but the pillars stay the same, real business operations producing real captures, ruthless niche-mastery so i'm always early, daily writing practice off the platform so the prose stays human, a tight network of real operators who can vouch for each other when the signal gets noisy
the bot-saturation problem makes the framework in this article more valuable, not less, the noisier the channel gets, the more the people doing the boring discipline work stand out, by accident as much as by design
earned permission
you read 4,000 words about a system that has no shortcut
i'm aware that's not the article most readers came for
most came for “the 5 ai prompts that grew my account 100x,” they're going to find that piece somewhere else, apply it, and 6 months from now wonder why their account looks the same as it did before
the people who do the boring version, the value tax, the daily rhythm, the tight micronetwork, the manual content loop, the niche-mastery, the read-and-write-off-x discipline, will be 12 months ahead of the prompt-shopping crowd, possibly more
i can have fun on x now, post the joke, drop the ragebait, do the engagement farming, none of which would have worked on day 30, all of which work now because i paid for the right to do them
you can pay for the same right
the system is on this page
the boring version is the version that works
How to build and scale a one-person business with AI:
