The setup
A senior operator in physical AI came to us with a familiar problem. She had two decades of operating experience in robotics and embodied systems, with the deep expertise and strong opinions that come from that kind of tenure. Her LinkedIn presence didn't show any of it. To anyone reading her feed, she looked like a generalist.
Her existing posts used broad AI and tech language. She wrote about industry trends, product launches she'd noticed, occasional reactions to news. The language was professional and the cadence was consistent. The problem was that nothing about it told a reader who she actually was or what she actually knew.
She'd been hedging on her real niche because she worried that going narrow would limit her audience. The logic seemed reasonable. More people are interested in "AI" than in "physical AI specifically." Going broad seemed safer.
The result was the opposite of safe. In a feed full of AI commentary, broad commentary disappears. The narrower the niche, the easier it is to be the obvious answer in that niche. She was being read as one more generic voice in a feed full of them. She could have been one of fewer than fifty operators globally who'd actually shipped physical AI products at scale.
What we did
The Expert Interview was where the work happened.
Ninety minutes, recorded, structured around a single line of questioning. What specifically have you learned from inside companies building physical AI products that someone reading about it from the outside wouldn't know?
The answers came in waves. She described the gap between what robotics demos look like and what they cost to actually deploy. She walked through supply chain decisions that determine product viability before any code is written. She got into the integration tradeoffs between custom hardware and off-the-shelf components, and the cases where one wins over the other. She explained why most physical AI startups die in the gap between technical demo and operational scale.
None of this had appeared in her LinkedIn content.
By the end of the interview, we had a list of more than two dozen specific positions she held. These weren't "AI is important" positions. They were operational positions that could only have come from someone who'd actually been inside the work.
We collected voice samples from her past writing. A few articles she'd written years earlier, internal memos she shared with us, transcripts from talks she'd given at industry conferences. The writing team studied them to calibrate against her actual voice, which turned out to be drier and more matter-of-fact than the LinkedIn version she'd been producing.
Then we rebuilt her content engine.
The output
Eight LinkedIn posts per month, ghostwritten in her voice, every one anchored in a specific operational position. The posts were structured to signal "this person has actually shipped physical AI products" within the first two sentences, usually through a specific reference to operational reality that someone outside the work wouldn't make.
Two long-form LinkedIn articles per quarter, each unpacking one of the deeper positions surfaced in the Expert Interview. The first article was about why most physical AI startups fail at deployment, with three specific failure modes drawn directly from her experience.
A monthly content calendar mapped each post to a position from the interview, ensuring we weren't repeating ourselves and weren't drifting back into the broad AI commentary she'd been producing before.
Voice calibration happened over the first four weeks. The writing team and the executive went through three rounds of refinement. They flagged phrases that sounded too generic, calibrated against past samples, and adjusted cadence. By the end of the fourth week, she told us she could no longer reliably tell which posts she'd written and which we'd ghostwritten.
What changed
The shift in her feed was visible within the first quarter.
The posts that performed best were the ones taking specific positions on physical AI deployment problems, exactly the territory she'd been hedging on before. The broad-AI posts she'd written previously had averaged moderate engagement. The narrow-position posts produced engagement that was significantly higher, and more importantly, came from a different audience.
Inbound connections shifted. Previously she'd been getting connection requests from broad "AI enthusiast" profiles, marketing people, students. After the repositioning, the connections coming in were operators at other physical AI companies, investors who specialize in hardware-software integration, and senior leaders at companies trying to hire someone exactly like her.
A handful of inbound business conversations started that hadn't been on her radar before. One was from an investor she'd been trying to get a meeting with for two years, who'd seen her cornerstone article and reached out directly.
She didn't grow her follower count dramatically. She didn't try to. The growth that mattered was qualitative. The audience returning to her feed was the audience she actually wanted there.
Why it worked
Narrowing was her competitive advantage. Broadening would have buried her.
Most senior operators try to cast a wide net on LinkedIn because they're worried about missing opportunities. Casting wide makes them invisible. The math is counterintuitive. In a category where ten thousand people are commenting on "AI," being one of fifty people commenting on "physical AI deployment specifically" is structurally better. The audience for "AI" is enormous. The audience for "physical AI deployment specifically" is the audience that matters.
The Expert Interview surfaced what she'd been hiding because she didn't realize she was hiding it. She'd assumed her operational experience was background context. We treated it as the foreground. That shift from background to foreground was the entire repositioning.
The voice work mattered, too. If we'd written the content but kept it in the broad AI register, the narrowing wouldn't have landed. Voice fidelity made the narrow positioning credible. The posts had to sound like an operator describing operational reality, not an outsider commenting on the field.
She still posts about broader AI topics occasionally. The difference is that those posts now reference her physical AI experience as the lens. Everything is anchored. Nothing floats.