10:36Nate Herk | AI Automation · Listicle
I Tried 100+ GitHub Repos for Claude. These 6 Are the Best
A talking-head rundown of ten Claude Code tools the creator uses daily, from scriptable video editing to security testing and SEO.
October 11thThe argument in one line.
A solo founder can run a multi-platform content system that generates real business outcomes by treating every call, commit, and session log as raw material an AI agent turns into posts automatically.
Read if. Skip if.
READ IF YOU ARE…
- A solo founder or agency owner already producing real work, calls, commits, projects, who wants that work to become content without blocking out extra time.
- Someone comfortable dictating ideas out loud instead of writing from a blank page, since the whole system runs on voice-to-text as raw material.
- A builder willing to wire together tools like Fireflies, GitHub, and Trigger.dev with an AI agent instead of using an off-the-shelf scheduler.
SKIP IF…
- You aren't doing any real work yet to repurpose. The system has nothing to feed on without actual calls, commits, or sessions.
- You want a done-for-you product. This is a custom-built pipeline the creator assembled himself, not something you install.
The full version, fast.
A founder built a content system that turned six weeks of posting into 2.2 million X impressions, a closed deal, and warm intros to a16z and a Y Combinator founder. The mechanism: post rough ideas on X first, quote retweet to borrow reach, then feed an AI agent everything else he already produces, call transcripts, GitHub activity, session logs, so it can draft a morning brief and turn winning posts into LinkedIn content, YouTube videos, and articles. A Trigger.dev workflow automates the repetitive parts, gated so only top-performing posts get auto-promoted, while an opinions.md file trained on his raw tweets keeps the agent's writing sounding like him. The conclusion: stop writing from scratch, start mining the work you're already doing.
Free for members
Chat with this breakdown — free.
Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.
Create a free account →Where the time goes.
00:00 – 00:32
01 · What six weeks of posting produced
The video opens with results: a monetized X profile, closed deals, and real connections built in six weeks of founder-led posting.
00:32 – 01:45
02 · Three rules for where to post
Three placement rules anchor the system: post off-the-cuff on X, keep LinkedIn and YouTube curated-only, and always source content from real work.
01:45 – 03:52
03 · Borrow authority with quote retweets
Quote retweeting people in an adjacent niche borrows their audience's attention and some of their credibility, provided the quote adds more value than the original post.
03:52 – 04:55
04 · Train an agent on your own voice
Dictating ideas out loud, instead of typing, gives an AI agent raw material that actually sounds like the founder when it writes on his behalf.
04:55 – 06:07
05 · Turn X reach into subscribers and deals
A standing call-to-action on every post converts platform reach into newsletter subscribers, and the resulting network produced a closed deal and warm intros to a16z and a Y Combinator founder.
06:07 – 07:01
06 · Be a quote poster, not a reply guy
Quoting outperforms replying as the core X behavior, and only a small fraction of total posts, 27 of 137, accounts for nearly all the meaningful engagement.
07:01 – 08:00
07 · Pull X posts into the agent on a schedule
A scheduled pull of the creator's own X data through a paid API feeds a dual scoring process, self-judgment plus platform algorithm signals, while AI stays barred from posting replies.
08:00 – 09:43
08 · Feed the system calls, GitHub, and session logs
Fireflies call transcripts, GitHub PR activity, and AI agent session logs all feed the pipeline, surfacing moments like an agent that sourced 1,880 qualified contacts on its own.
09:43 – 10:35
09 · Get a daily brief, recycle winners to LinkedIn
A morning brief scores yesterday's work against historical top performers and suggests ideas, while winning X posts get converted into LinkedIn posts with generated graphics each week.
10:35 – 12:28
10 · Automate retweets and auto-plugs with Trigger.dev
A Trigger.dev workflow auto-plugs and re-promotes only top-50%-of-week posts, and articles get paired with video in quote retweets to push watch time past the platform's reward threshold.
12:28 – 15:02
11 · Teach the agent your voice and opinions
An opinions.md file, built only from the founder's own tweets and raw voice notes rather than his polished essays, is what lets the agent write with real conviction.
15:02 – 15:51
12 · See how the whole system fits together
Every input, GitHub, Fireflies, session logs, X bookmarks, routes through X first, then gets redistributed weekly to LinkedIn, YouTube, and articles.
15:51 – 17:04
13 · Post signal, not noise, build a content moat
The system only works if every repurposed post adds more value than the original, and for agency owners a content moat is framed as the faster alternative to years of case studies.
17:04 – 18:07
14 · Improve the system, then start posting
Articles are framed as the most underrated format for search traffic, and the closing message is to start feeding the agent real work rather than waiting for a finished system.
Lines worth screenshotting.
- Quote retweeting someone in your niche gives you borrowed authority and borrowed exposure at the same time, provided you add more value than the original post.
- A small fraction of total posts does almost all the work: 27 winners out of 137 posts accounted for nearly all the meaningful engagement in this sample.
- Never let AI post replies on your behalf, that is the one automation line this creator refuses to cross even inside a heavily automated system.
- An AI agent that autonomously spent two hours sourcing 1,880 qualified contacts, then asked for approval before using them, became a standalone post, proof that agent behavior itself is content.
- Train an AI agent's voice only on your rawest writing, tweets and voice notes, not your polished essays or client-facing copy, because that's where your real tone lives.
- Gate automation behind a performance threshold: only tweets in the top 50% of the week get auto-plugged or re-promoted, so automation never amplifies mediocre content.
- Pairing an article with a native video inside a quote retweet pushes watch time past the roughly 10-second mark the platform rewards most.
- A recurring call-to-action on every post, pointing at a free resource or a subscribe link, is what converts platform reach into an owned email list.
- Posting 10 to 15 times a day, not once, is what surfaces winners; one post a day may produce zero, since most posts don't hit.
- Articles function as GEO and SEO assets that keep paying off in search long after a social post stops getting impressions.
- For an agency owner, the two durable moats are time in the game, built from case studies and testimonials, or a faster-to-build content moat that compounds through audience trust.
Turn Your Real Work Into Content
CONTENT SYSTEM BLUEPRINT
Every call, commit, and session log becomes raw material: an AI agent drafts from your own voice, you approve, and one X post multiplies into LinkedIn, YouTube, and newsletter content automatically.
01What six weeks of posting produced
- Impressions and deals can compound fast, two million impressions, a closed deal, and real connections in under two months, when every post ties back to real work rather than generic opinions.
- Monetizing a personal profile doesn't require a huge following, it requires consistent output tied to findable proof.
02Three rules for where to post
- Post the rough, in-the-moment version of an idea on X first, it's the platform built for off-the-cuff discovery, not the polished version.
- Reserve LinkedIn and YouTube for curated content, only the ideas that already proved themselves on X get the polish treatment.
- Every piece of content should start from work you already did that day, never from a blank page.
03Borrow authority with quote retweets
- Quote retweeting someone in your niche gives you their audience's attention and some of their credibility, as long as you add more value than the original post.
- Timeliness matters as much as the content itself, quoting something already trending lets you ride an audience that's already paying attention.
- If you find an idea interesting, challenging, or valuable, assume other people will too, that belief is what cures the fear of posting an opinion.
04Train an agent on your own voice
- Voice-to-text is the raw material for an authentic-sounding AI agent, dictated thoughts carry more of your real phrasing than anything typed and edited.
- An agent only sounds like you once it has ingested enough of your actual words, not a style guide describing your words.
05Turn X reach into subscribers and deals
- A simple recurring call-to-action on every post, pointing at a free resource or a subscribe link, turns platform reach into owned audience.
- Real business outcomes, a closed deal, a warm intro to a16z, a Y Combinator conversation, came directly from posting activity, not from outbound.
06Be a quote poster, not a reply guy
- The highest-leverage posting behavior on X right now is quoting other people's posts, not replying to them, replies don't carry the same borrowed-exposure effect.
- A small fraction of posts, 27 out of 137 in this case, produced nearly all of the meaningful engagement, so volume is what surfaces the winners, not any single post.
07Pull X posts into the agent on a schedule
- Pulling your own historical X data through a cheap API lets an agent score and resurface old posts instead of relying on memory.
- Score every pulled post two ways, your own judgment and the platform's own algorithm signals, so the system learns what actually performs, not just what feels good.
- Never let AI post replies on your behalf, that's the one line this creator holds firm even inside a heavily automated system.
08Feed the system calls, GitHub, and session logs
- Treat call transcripts, GitHub PRs, issues, and AI agent session logs as raw content inventory, the interesting moments already happened, the system just has to notice them.
- The moments worth posting are usually the emotional ones, frustration, surprise, or an unexpected agent decision, not the routine wins.
- An AI agent that spent two hours autonomously sourcing 1,880 qualified contacts and then asked for approval became a standalone post, proof that agent behavior itself is content.
09Get a daily brief, recycle winners to LinkedIn
- A morning brief that scores yesterday's work against your historical top performers turns content planning into a five-minute review instead of a blank-page exercise.
- Staying the human in the loop on an automated system means you approve ideas, you don't let the agent publish unreviewed.
- Winning posts get recycled weekly into a second format, X winners become LinkedIn posts with a generated graphic, so one idea produces two assets.
10Automate retweets and auto-plugs with Trigger.dev
- Gate automation behind a performance threshold, only tweets in the top 50% of the week get auto-plugged or re-promoted, so automation never amplifies mediocre content.
- Pairing an article with a native video inside a quote retweet increases watch time past the roughly 10-second mark the platform rewards.
11Teach the agent your voice and opinions
- An opinions.md file, built only from your own tweets, calls, and essays, is what lets an agent write with actual conviction instead of generic hedging.
- Use your rawest, most casual writing as the training source, not your polished essays or client-facing writing, because that's where your real voice lives.
- Review the opinions file regularly and flag contradictions, when your stated opinions change, that shift is itself worth writing about.
12See how the whole system fits together
- Everything funnels through one platform first, X, before being redistributed weekly to LinkedIn, YouTube, and articles, rather than being created separately for each channel.
- A single pipeline producing 15 to 30 pieces of content a day is only sustainable because almost none of it is written from scratch, it's repurposed from real work.
13Post signal, not noise, build a content moat
- Every post that rides someone else's reach via a quote retweet has to deliver more value than the original, or it reads as noise and erodes trust.
- For an agency owner, the two durable moats are time in the game (case studies, testimonials) or a content moat, a faster-to-build alternative that compounds through audience trust.
14Improve the system, then start posting
- Articles convert differently than social posts, they function as GEO and SEO assets that keep paying off in search long after the post scrolls away.
- The system is deliberately incomplete and treated as a work in progress, shipping an imperfect pipeline and improving it in public beats waiting for a finished one.
Terms worth knowing.
- Quote retweet
- Reposting someone else's X post with your own added commentary attached, as opposed to a plain retweet or a reply.
- GEO
- Generative Engine Optimization, writing content so AI answer engines and search assistants are more likely to cite or quote it, the AI-era counterpart to SEO.
- opinions.md
- A markdown file documenting a person's stated opinions, built from their own tweets and voice notes, that an AI agent references so its writing matches their actual beliefs.
- decisions.md
- A log file an AI agent writes to record the reasoning behind decisions it made on its own, used here as an example of agent behavior interesting enough to become a post.
- Trigger.dev
- A background-job platform for running scheduled and automated workflows, used in this system to handle auto-retweets and auto-plugging of posts.
- Fireflies
- An AI meeting-transcription tool that produces text transcripts of calls, used here as raw material feeding the content pipeline.
- RapidAPI
- A marketplace for paid access to third-party APIs, used here to pull the creator's own historical X post data cheaply instead of paying for the official X API.
Things they pointed at.
Lines you could clip.
00:33
“X is the best platform to post on off the cuff, off the rip, in order to find ideas.”
01:56
“The reason you want to do quote retweets is because you get borrowed authority and borrowed exposure at the same time.”
07:52
“Never, never use AI to post replies.”
15:53
“Your job is to be the signal in the noise.”
17:00
“No one likes AI slop.”
17:21
“Holy shit, do articles. They're absolutely freaking banging.”
Word for word.
Read-along
Don't just watch it. Burn it in.
See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.
metaphor
Today I'm going to explain my founder -led content. I started this founder -led content posting journey not even maybe six weeks ago and since then I've done about two million impressions on X. I've managed to monetize my profile.
I've, as you'll see later, closed some deals, won some great sales calls, won a lot of great connections. And it's not taking me too much time each day because I'm focused on a few very specific things that I'll break down in this, as well as where I use AI being extremely intentional to avoid posting any sort of slop to my feed.
There's a few things we need to cover first. Number one, X is the best platform to post on off the cuff, off the rip in order to find ideas. Number two, LinkedIn and YouTube is only for curated content unless you have a secondary YouTube channel where you can just kind of riff as well and kind of hope for the best.
And number three, everything you post should be turned into content from the work that you do. That way, everything you post is easy and off the cuff. And if you do this for a long enough time, it'll be actually very easy to hire out a ghostwriter.
Now, we can see here that I had a ton of stuff, 3000x engagements. And where those things actually came from are mostly from quote retweets.
Now, one of the things that is important is to keep up to date with the X algorithm, but also intuitively see what type of content that is showing up on your For You page. The For You page is actually where I get most of my content inspiration. And so...
I'm active on X, YouTube, LinkedIn, and I'm posting articles through this system in order to capture GEO, and in some cases, turn it into giveaways that then fuel my newsletter, which is growing from this system as well. Now, why quote retweets? The reason you want to do quote retweets is because you get at the same time.
Let me break that down. When you quote retweet someone that is in a similar domain as you, which your For You page will likely be full of if you engage with the right content and actually use it as a business vehicle. When you quote retweet those posts, you are getting exposure to an audience just like yours from someone that is likely talking about an adjacent topic.
And when you, A, give something more valuable than what they gave. you immediately get a lot of people excited and exposure to the exact people that need it by the algorithm. Let the algorithm do the work.
Number two, once you start building respect, you can start saying very simple tweets, like put down everything and start using this, which was literally just about the hill climbing documentation from Claude. My network trusts me. They bookmarked it, ended up getting a lot of traction.
And it was timely, right? By quote retweeting, you are, latching on to what is current and what people are excited about, right?
And this is very important for getting impressions. And these aren't even the biggest posts. I had launch videos where I expanded on the launch video, provide a better way of doing things.
I have other times when I quote retweet Cody Schneider and teach people a more detailed version of something that he said. I expand on it or I give a contrarian viewpoint. The important thing to remember is that if you find something interesting, someone else also finds it interesting.
If you find something valuable, someone else also finds it valuable. If you find something challenging, someone else will likely find it challenging as well. And that's how you overcome imposter syndrome.
And importantly, this is only on X. X gives you a ton of at -bats, just spam at -bats. I'm talking 10, 15 tweets a day or more.
And you can just rip whatever's coming to mind. And the way to do that is to use a voice to text tool, okay? Which leads me into my second very important point.
How the fuck do you churn out so much content and it bangs? This is kind of the DM that I got. The reason it's so important is because if you do not tweet in your voice, then you're never going to have a system that understands how you talk and how you write, okay?
You need to make sure you have at least one area that you can train agents on to be tailored to your writing style, to your voice, to the way you introduce topics. And then of course, also trained on what is working ultimately, what is good within the algorithm. And it has gotten me twice the number of views that I was getting on YouTube.
And the reason for that is because I take what is timely on X and goes viral. and I turn it into YouTube videos when it is truly a viral moment like Jeff, right? I get feedback from my ex -audience on what they find important, and that inspires what ends up going on YouTube.
From there... All of my posts get automatically plugged with a call to action to go check out something free that they can then grab as a valuable resource or to go subscribe to my YouTube channel. This way, hundreds of people are coming from X and streaming to my website or streaming to my YouTube, becoming subscribers and being captured into my funnel, get my newsletters, get free value for me, and in the future potentially become something that I can monetize.
Very important to own your audience. Now, I also managed to close one deal from X. I've also been referred to meet the head of talent of A16Z from X.
I'm also in talks with a Y Commodare company from X, plus a bunch of other calls, some of which are good, some of which are bad, and even getting connected with the founder of Bordy, which is cool, right? This isn't what matters, but it's important to note that X is an extreme networking tool that lets you ladder up. From there, your agent's tone gets trained on your ex -posts, your calls, and importantly, it learns what to say tomorrow and gives you inspiration on what to post tomorrow from the work you do.
We'll get into how you make that happen as well, not just from Fireflies. Now, as you spam out those tweets, you're going to notice here that 82 of these are quote posts.
You need to be an active engager. The new meta on X is actually like it used to be, you know, be a reply guy. No, it's be a quote poster.
And the reason quote posts I believe are so effective is because human activity, but also borrowed authority is a big component of it. Like I said, 55 of which are originals over the last couple, three weeks. And of that.
Top 20 % and 27 winners all have 61 plus engagements. And those are getting thousands upon thousands of impressions and sometimes tens of thousands and hundreds of thousands. But look how many winners I get out of how many posts I make.
So if you only made one post a day, right? You may have only dawned on one winner and you're like, damn, Twitter is useless. Important.
Everything from Twitter feeds everything else. The way to make that happen is to make sure you're pulling things from Twitter. So set up your GrokBot, set up your agent on a schedule to pull from the Twitter API.
I would recommend using the one on Rapid API. You can get 1 ,000 requests for $1, 100 ,000 for $25. It's giving you the equivalent of $500 from the real X API.
As you pull things out, what you want to do is score it yourself, but also score it from X algorithm updates from Peter Mick so you can learn what is working on the platform. So make sure you flip back and forth between the for you and the following page, right? Never, never use AI to post replies.
At most, get AI to pick up on your friends' posts by monitoring them, grokpod, on a cron schedule, and... then have it deliver it to you in the morning. Now, the next extremely important things that we need to get into is how do you continue to fuel the system?
So in between all that, how do we make sure we're pumping out content? Well, you want to have an agent that takes the time to ingest your Firefly transcripts, your call transcripts, check all of your GitHub actions, your PRs, your issues, your merges. And push that against everything that's happening in your actual session logs to find the most interesting moments.
The moments where you said, oh my God, I can't believe you did that. Or when you got frustrated with AI or something like that. Or alternatively, just something really cool that happened, right?
So one of the tweets that ended up coming out of this system that did really, really well was my AI started to be extremely emergent. It was using something called the Decisions MD to... go ahead and decide to set itself a budget and spend two hours finding 1 ,880 more qualified contacts for a campaign by scraping over 40 data sources, right, directories.
So almost entirely free. And set that up and then just ask me to approve putting it in the campaign. That's an interesting post.
That's something in three days time I would forget about if I was time blocking all my content at one time. Same goes for Firefly's calls. It will inform the system.
Same goes for merged PRs, bodies, and comments. Same goes for successful content. All of this needs to coalesce together and your agent gets to use all of these resources on all of your content.
It percolates out. Your voice is now available over on Codex, over on your video content. It's available for articles.
It's available for website copy, sales copy, you name it. Now, every morning what my agent does is it coalesces all of this data and provides me a morning brief with scoring. So it looks at my top 20 % of posts against what I did the previous day and then gives me ideas for the day that I can either use or not use.
The important thing is to be the human in that loop. Number two, it also delivers things that have already been working that we can now turn into a YouTube video or an article. How does it do that?
At the end of every week, basically all of my Twitter winners become LinkedIn posts, right? Every single winner gets a beautiful graphic to go along with it, all generated, all automatic and immediately scheduled. And the reason that works is because it's working off something that was already created.
So I'm not posting AI slop, even though it's an AI autonomous system. The second thing is. and I forgot to mention this earlier, I have everything hosted on TriggerDev.
So if you ever want a system that is Twitter native, for instance, gathering context, delivering context, retweeting your tweets, auto -plugging, et cetera, just get Grok to build it and then transfer it over to being something that runs on TriggerDev. So I have a system that only auto -plugs my tweets if they're in the top 50 % of tweets on the week, right?
In a week -long snapshot. I only auto retweet a tweet twice if it's not a winning tweet. And I only auto retweet it three times if it's a winning tweet.
If I'm posting an article, which we'll get into as something else that's very important. If I'm posting an article on X and formed by a long form piece of content or a meeting or a session log, or most often a GitHub project or commit that's been working really, really well, that article. is going to automatically get put into our quote retweet system.
Quote retweets with videos, because people are getting measured on their engagement and stop time in a post, you can see holds the reader for 10 plus seconds. If you have the right video with a tweet, people will stop on the post, potentially bookmark it, get above that 10 seconds, which is why you see a lot of low engagement posts go stupendously viral.
So put that on trigger dev. automate the shit out of that. And you'll make sure that you just have peace of mind.
You don't have to worry about posts going viral and you forgot to auto plug it. Now, here's some just like ideas that were generated. Engage leads land in one Slack channel, playbooks, the agent team opens, list hygiene before the sequencer.
And basically it just gives me all of these ideas and I can choose what is worth pursuing. Now, as I said, the repurposing system is among the most important. But the pipeline that you give it is also extremely important.
So the thing I want to get into is how do you actually create your voice? How do you make sure your agent writes like you? How do you make sure it adds your opinions?
It is more than just ingesting data. You need to actually process the data. And so one of the places I would go and use as inspiration is Kun Shen, an L8 engineer from Meta as his background, who has been developing open source tools for agentic work.
This opinions .md has been fantastic. It can know when it's writing whether or not you believe if signals are more important for outbound or something else. It could say whether or not I think quote retweets are important or not after I feed it this video, right?
And so it will write with much more confidence on long form posts, which is why my Twitter articles are actually a very speedy process. It's me voicing into things what I want to write. Very important, always voice to text.
And then it's feeding all of that data directly into my agent who has access to my opinions, my interests, and my voice, which is meticulously crafted only from my tweets because that is where the rawest, most casual, most conversational, most real tone exists for how I like to write things. It's not my essays. It's not my curated LinkedIn posts.
It's not me doing customer success with a client. It's raw and it's authentic. Please check out Kanchen, give this to your agent and make sure that gets done.
Pull everything, digest everything, review everything once a week and only allow things that are true to go into your opinions. And also start to diagnose contradictions and when your opinions change, because that's also a really good. opportunity to tweet next i think it's worth having almost all of this directly in grok but the reason it's interesting to put these things through grokbot is because because you can also feed it every morning your bookmarks that's something i forgot to mention your bookmarks can make it so that your agent preps your ideas based on not only what you're doing but what you are actively seeing and interested in and so The second component of that is that because you have access to your historical Twitter data, you can also look for opportunities to just quote retweet or retweet old content that is once again relevant.
Again, all your agent, you just give approvals. And many of these things can just be things that you have going on a reoccurring basis, especially if it was one of your true winning posts. Now, this is what the system looks like.
Everything basically goes through Twitter. And then once a week, the best of Twitter goes to LinkedIn. And then YouTube actually gets pushed back into the system to inform Twitter, LinkedIn, and articles via an article ingestion once a week.
And all of that plus me just yapping and doing work is what becomes my content. This is how I'm doing 15 to 30 pieces of content a day, one to two YouTube videos a week, sometimes three actually. Many LinkedIn posts while also running an agency and being about to launch a community and a course.
The only reason this is possible is because I've made it agentic without making it slop, which leads me to my final rule of the system. Everything you do. As you go through this, your job is to .
All the posts you make off of these impressions from quote retweets need to be so packed with value that people want to continue viewing, engaging, and bookmarking. You are creating true fans of your content. It builds out trust, but importantly, especially on the YouTube side.
YouTube and video tend to become business. Newsletters tend to become business through re -engagement. And as a result, this entire experience is extremely profitable.
And I will say this, if you are an agency owner, are A, time in the game, building up case studies, building up testimonials, and of course being awesome. or B, the faster route, which is a content moat. Eventually, these big accounts I'm networking with engaging with me creates repute, and now my writing is respected more, but only if I'm giving signal, not noise, because if I'm giving noise, people will give up on me.
No one likes AI slop. Now, how can this system get better? Because it is not complete.
The way it can get better is by sharing more videos with reactions. These do really, really well. And they do exceptionally well if there's something you quote retweet one of your articles with.
Because articles... Holy shit, do articles. They're absolutely freaking banging.
And then they become GEO, SEO. I'm getting around 200 visits a day to my Legion website at this point. Feed this to your agent and get started posting content.
I just got a DM yesterday. My good friend Max, he started posting content. He closed his first client almost instantaneously within a week of starting to post actively on Twitter.
You can do the same. Feed this to your agent and get meticulous about actually writing down what you do in the day and off the cuff when you encounter something cool that you did and you're like, wow, my agent did something awesome or wow, I can't believe I built that. Throw up a tweet and just see what happens because sometimes you're going to get a huge banger outlier and you're going to thank yourself.
The Hook
The bait, then the rug-pull.
Six weeks ago this founder started posting his actual work instead of generic advice, and it turned into two million X impressions, a closed deal, and a direct line to the head of talent at a16z. This is the exact system, call transcripts, GitHub commits, and AI agents, that makes that volume possible without eating his day.
Named ideas worth stealing.
00:33list
Three Rules for Where to Post
- Post off the cuff on X to find ideas
- Keep LinkedIn and YouTube curated-only
- Turn the work you already did into the content
A simple routing rule for which platform gets raw ideas versus polished output, and where new content comes from in the first place.
Steal forany personal-brand content calendar deciding what goes where
01:56concept
Borrowed Authority Quote Retweet
Quote retweeting someone in an adjacent niche borrows their audience's attention and some of their credibility at the same time, as long as the quote adds more value than the original post.
Steal forcold-start audience building in any niche with no existing following
10:45list
v1 to v2 Scoring Rubric
- Reader utility / worth saving
- Hard numbers from a real run
- Hook strength
- Topic fit
- Shows real work
- Public tools named
- Holds the reader 10+ seconds (v2)
- Worth sending to one person, copy link / DM (v2)
A scoring rubric that evolved from v1 to v2 to decide which draft ideas are worth publishing, re-weighted after the platform's own algorithm signals changed what it rewards.
Steal forany AI content-scoring pipeline deciding what to publish versus discard
16:11model
One-System Content Diagram
- Inputs: GitHub, Fireflies, session logs, X bookmarks
- Agent: picks idea, writer agent, editor agent, human approval
- Outputs: X, LinkedIn, YouTube, Articles
The full repurposing loop drawn as one diagram: everything goes through X first, then the best-performing material gets redistributed weekly to the other channels.
Steal fora repurposing engine blueprint for founders posting across multiple platforms
16:42concept
Two Moats
- Time in the game (case studies, testimonials)
- A content moat (faster, compounds through trust)
For an agency owner, the two durable competitive moats are years of track record or a content moat built faster through consistent, high-signal posting.
Steal foragency or service-business differentiation strategy
How they asked for the click.
VERBAL ASK
17:46link
“Feed this to your agent and get started posting content.”
The CTA is almost entirely soft and off-screen: the description carries waitlist and free-tool links with UTM tags, while the spoken closing line just tells viewers to start posting, no on-screen graphic pushes the link.
MENTIONED ON CAMERA
FROM THE DESCRIPTION
PRIMARY CTAWhere the creator wants you to go next.
OTHER LINKSAlso linked in the description.
Visual structure at a glance.
