The copycat fear in building in public is real but overrated. Anyone can screenshot my numbers. Nobody can copy the 25 years of watching campaigns and clients that tell me which numbers matter. Transparency spreads the what, never the why.
https://t.co/YEyjE4yTSU
I don’t often talk about the downsides of building in public:
> I’m super transparent about our roadmap. competitors have launched similar features suspiciously around the same time (on many occasions)
> I wear my emotions on my sleeve. I’m outspoken. I’m competitive. that’s ruffled some feathers, and we have definitely lost (overly-sensitive) customers bc of it
> when you normalize sharing your revenue / growth / launches … there’s no faking that. you can’t hide a bad month
> it can be time consuming. I have 130+ employees and a never-ending to-do list of priorities. the extra time to create and share content (like this post) adds up
BUT there are many more upsides than downsides (IMO):
distribution: I have 150k subscribers on my personal newsletter and another 150k across socials. that’s pretty decent distribution to reach people and share updates
talent: almost all of our employees come inbound. most of them have already come across my newsletter or content on social and feel familiar with the company and how I operate
revenue: TIME is one of our biggest enterprise users. their CTO signed up to my newsletter, read it for six months, then convinced the org to move their newsletters to @beehiiv
fundraising: when we raised our Series A, I didn’t do a single coffee chat (or even make a deck). investors had been reading our updates for months and they already knew the numbers
//
there’s definitely a vulnerability to building in public that makes it more relatable. if it’s always positive news and revenue milestones, no one wants that. they want what’s actually happening behind the scenes
I’ve heard both sides of the build in public debate — what do you think?
Building the tool your own agency runs on is a cheat code most SaaS founders don't have. Every bad draft costs me a client, so the fix ships that week. You can't fake that urgency when your only user is a beta tester in a Slack channel.
We don't run a humanizer over generic output and hope. We calibrate on 5 of your real posts first, so the variance is yours: your sentence swings, your odd phrasing. A separate agent checks facts so nothing gets "humanized" into being wrong. https://t.co/4VbE2z4GzP
https://t.co/HZqMeeVpP1
Editors: what's the first AI tell you catch before you can even explain why? For me it's the sentences that are all the same length. Everything sits at a flat 15 words and the rhythm goes dead.
The dirty secret in the humanizer arms race: most tools "pass" detectors by making the text worse. Choppier rhythm, synonym stuffing, and they quietly change your facts while they're at it. You end up with slop that reads sloppy on purpose. That's not a fix.
https://t.co/SebWpquqZh
The use of AI writing tools is everywhere from LinkedIn to Substack and just about everywhere in between. So it's no surprise that AI users are looking for ways to disguise their writing in hopes to bypass AI detectors like GPTZero and Pangram.
To see if these humanizer actually work, I tried several different ones to see the results. As it turns out, the writing ended up sounding more like AI slop than what the AI writing tool churned out in the first place.
Clear sentences came back stuffed with bizarre synonyms and the rhythm became choppy and unnatural. In some cases, the tools changed facts or stripped away important context. The results sounded anything but human.
The reason "humanize AI text" is the top search in this whole category: people are trying to sand off tells instead of writing something worth reading. You can defeat every detector and still bore everyone. Passing Pangram is not the same as being good.
On Upwork you get judged on turnaround and accuracy, not effort. A separate agent cites and fact-checks every draft, so you hand the client a piece with sources they can click. That's the invoice line they don't argue with. https://t.co/RKaoi1pyTl
This is right, and it kills the keyword-stuffing hustle. Clients paying a premium don't want a note that says "AI-assisted." They want to know who chose the sources, who checked them, and who answers when a claim breaks.
https://t.co/h2EnXazgEU
I think you’ve landed on a question I’ve also been asking: what does “wrote” mean now?
If a client is paying a writer $250/hr they absolutely deserve transparency about the service they’re buying & the standards writer is accountable for.
But I don’t think human authorship can be reduced to who physically generated every sentence.
I’ve spent decades working with writers, agencies, designers, strategists & creative teams. I’ve ghostwritten for executives & I’ve hired writers who became extensions of my brain. The best work was always collaborative.
AI adds another collaborator to that process.
I can spend hrs developing an idea, challenging it, researching, going back & forth with AI, rejecting directions, adding my experience, rewriting, editing & deciding what ultimately gets published. I’m not comfortable calling that “content I never wrote.”
And I definitely don’t believe AI output is inherently mediocre any more than I believe human output is inherently good.
To me the standard is still the same:
Who did the thinking? Who provided the judgment? Who owns the final work? Most important… does it deliver value to the audience?
I also believe writers & creators should be part of defining what those standards are & how they should evolve
We built https://t.co/aY3heyDwgp so the research and fact-check ride along in a separate agent, cited and checked. You sell the brief and the accountability. The words come standard.
The client saying "AI could write this for free" is telling you exactly what to charge for. Not the words. The brief that says what to write, the research behind it, and the name on the accountability when it's wrong. That part was never free.
We feed 5 of your existing posts in first, so the paid drafts read like the free ones your subscribers already opened. A separate agent fact-checks and cites before it ever hits your queue. 7-day trial. https://t.co/TcZ0Jx6z94
Substack writers: when did you turn on paid, and what actually made subscribers convert? A specific post? A cadence? Being loud about the button? Curious what moved the needle vs what you assumed would.
The "write what you'd enjoy reading" advice is right but incomplete. You still have to sound like a person while doing it. Plenty of writers enjoy their own drafts and still read like a settings menu. Enjoyment isn't a voice check.
https://t.co/2zlmqkeRde
It works. Highly recommend. Best way to get the audience/followers you want. Even better, don’t “try” to do anything. Just write the kind of content you’d personally want to consume if you were the one reading it, or that you at least genuinely enjoy sharing. I’ve made a lot of great friends this way and gotten exposure to opportunities I was never credentialed for. I’ve been able to sit in rooms with people much smarter than me and, to my surprise, they actually wanted to hear what I had to say. That was the epiphany for me that I needed to shift my focus away from building a larger audience and instead focus on quality over quantity. You can be Walmart. That’s cool. We need Walmarts. Just realize that you’re going to have to deal with Walmart people.
Social media also removes a lot of the communication barriers that present themselves in real life and gives you an immediate, deep insight into how someone thinks. Not selectively pruning your audience is a big mistake you will regret.
https://t.co/geL2BtcFzt
Everyone quotes the "only 5% monetize" stat like it proves paid Substack is a scam. It proves most people paywall before they sound like anyone. Readers don't pay for a topic. They pay for a voice they can't get from the other 40 newsletters on the same topic.