Do not index
Should you stop using AI for your LinkedIn content the way Steven Bartlett's team just did? Not entirely, but the reason his team's decision worked is the part most founders will miss if they only copy the surface of it.
FlightStory, Bartlett's company, stopped using AI to write his LinkedIn posts after automated text became visible enough to the audience that it started costing engagement. His posts are now fully handwritten, and the team leaves occasional spelling mistakes uncorrected on purpose, to signal that the writing came from a real process rather than a generation pipeline. FlightStory director of business development Christiana Brenton framed the shift this way: "AI mistakes are really noticeable. When the world wobbles to the left, opportunity lies to the right. What will stand out on LinkedIn now more than ever? Real words written by a human." The result, per reporting on the change, was significantly better reach and response than the AI-assisted version of the same account.
The mistake would be reading this as leave your typos in. That is the least useful part of the story. The useful part is that a team with Bartlett's resources, meaning full access to every AI writing tool on the market, looked at the data on their own account and chose to remove AI from the thinking layer entirely rather than manage around it. That is a judgment call about where the line sits, made by people who could afford the best tools and decided the tools were costing them more than they were saving.
I call this the Fingerprint Rule: an audience can tell the difference between an unpolished byproduct of a real process and a manufactured signal of one. A typo that survives because a founder wrote a post at 11pm and hit publish reads as real. A typo inserted deliberately to fake that same effect is its own kind of performance, and audiences already primed to distrust AI-flavored writing are exactly the audiences most likely to sense when authenticity itself is being staged. FlightStory's imperfections work because they are downstream of a real handwritten process, not because imperfection is inherently trustworthy on its own.
This is directly relevant for founders running personal-brand content on LinkedIn, particularly those past the point of a few thousand followers where every post gets read as a signal about how they run their company, not just what they think. It is also relevant for the ghostwriters and ghostwriting agencies who work with those founders and have to decide, account by account, how much AI belongs in the process. It is not relevant for founders who genuinely have no time to be a source for their own content at all. If a founder cannot give a writer thirty minutes a week of real opinions, examples, and specifics to work from, the Fingerprint Rule does not apply, because there is no real process underneath the post to signal. That founder needs a different service model built around structured interviews and heavier ghostwriting, not a lighter AI-assist workflow.
What actually changed at FlightStory
The operational shift is smaller than the headline suggests. FlightStory did not walk away from a content team or a system. It walked away from letting AI touch the sentence-level construction of Bartlett's posts, while presumably keeping other tools for research, scheduling, and formatting. That is consistent with how the founder positioning conversation has been trending all year: a founder's own reasoning and specific point of view are the part of the process that cannot be outsourced to a model without the audience noticing, even when the model is good.
Why copying the typos will not work
Any agency tempted to instruct writers to insert deliberate errors into client posts is solving for the wrong variable. The typo was never the mechanism. The mechanism was a founder's real, specific, occasionally messy thinking making it onto the page without an AI layer smoothing it into something generic. Fake imperfection without that underlying process is a costume, and it tends to read as one within a few posts, especially to an audience that now includes people actively looking for AI slop signals to report.
The size of the company here is the part worth sitting with. FlightStory is not a two person shop without options. It has a full content operation and access to every AI writing tool worth using, and it still concluded that the model doing the thinking cost more in reach and trust than it saved in time. That is a stronger data point for a founder weighing this decision than almost any advice a ghostwriter could offer secondhand, because it comes from a team with nothing to gain from faking the result.
What Bartlett's team actually demonstrated is a pricing and process decision available to any founder with real opinions and thirty minutes a week, not a resource-intensive luxury reserved for companies with production budgets. The founders who take the underlying lesson, meaning that judgment has to stay human even when everything else gets faster, are the ones whose LinkedIn presence keeps compounding in a feed that is actively filtering for exactly that signal. The founders who take only the surface lesson, meaning leave in a typo, are optimizing for a signal they never actually earned.
