Natalie Barbu Is Building AI Worth Keeping

Natalie Barbu has been building in public since before the phrase had a name. She started a YouTube channel in high school in 2011, went full time as a creator in 2019, and right now an audience of around 300,000 on YouTube, plus another 15,000 on LinkedIn, watches her build Rella pretty much in real time.

An audience that size usually comes from feeding it. Posting constantly, chasing whatever the feed rewards this week. Natalie spent more than a decade learning exactly how that game works. Which is why the way she built Rella is a little surprising: it runs on the opposite instinct. Rella, the content planning and project management tool she started with three co-founders, is the product she needed back when she was running her own creator business on a mess of disconnected tools, and she is careful that nothing goes into it just to have shipped something. The proof is the year she spent not building the one thing everyone else was racing to add.

Where Rella came from

Natalie went full time as a creator in 2019 and ran a small agency on the side, helping creators and brands with their social strategy. That agency is where Rella started, though not as a plan so much as a frustration. She was holding her whole workflow together with a project manager, a separate scheduler, and a pile of spreadsheets, none of it talking to each other. "Everything felt so disconnected, and I knew there needed to be a better system for people who work in social media."

She was her own first customer, which tends to produce a really specific kind of product taste. From the start, Natalie wanted Rella to feel less like another piece of software and more like something on your side.

A tool that behaves like a teammate instead of a feature. That instinct sat in the company for three years before it got its real test.

Why Rella waited a year to launch AI

When ChatGPT put generative AI in front of everyone, most software companies added an AI feature within weeks. Adding one was the obvious move, and Natalie has never had much interest in the obvious move for its own sake. Only in whether it actually made Rella better. Ella, the company's in-house assistant, didn't launch until July 2025, about a year after adding AI became the reflex everywhere else.

"We didn't announce Ella until a year after everyone started adding AI immediately," she says. "It really was because I was like, we need to make this feel intentional. I don't want it to be this novelty AI that's cool, you use it once and then never use again."

Waiting, for her, was never the same as sitting still. "We actually took a while to observe how people were using AI in their systems and what would actually help them," she says. While other companies shipped their version and moved on, her team spent the year watching how people used AI inside real workflows, and how they leaned on Rella and the rest of their stack. The idea was to build for the thing people come back to, not the thing that looks good in a launch post.

She had a really specific picture of the thing she was refusing to build. "People would pay ten dollars to look like a 90s yearbook photo, because that was cool with AI at the time." Then the question she actually cared about: "How many repeat customers did that company have?"

And the detail that makes the whole thing click: no one had even asked for it. Rella's habit is building what customers request, and this was the opposite. "AI was not something anyone was requesting. No one asked for a tool like Rella." She built it anyway, betting that a version that actually understood a customer's brand voice and their analytics, and could do the work instead of just suggesting it, would earn a place a novelty never could. Ella connects a user's accounts and a persona, reads their past posts and captions and performance, and learns the brand's voice before it drafts a single thing. Then it writes captions, scripts, and a content plan that sound like the person using it, not like a tool guessing at a trend.

The line she uses to draw the distinction is the whole thesis of the company: "You can easily replace a feature. You can't easily replace a teammate."

From founder mode to CEO mode

The care Natalie puts into what Rella ships, she's now learning to put into how she runs the company. She's in the middle of that part now, and it's harder.

Rella has grown into the pressure. A thousand new users a month, a number that used to feel impossible, is now the floor. "A year ago that was like, holy cow, biggest thing ever," she says.

"Now a thousand people a month is kind of baseline." The team is eight people and hiring a ninth, a performance marketing manager to bring a funnel in-house that a paid ads agency currently runs across Meta and TikTok. The hire is meant to own the whole funnel and run acquisition experiments at a scale nobody on the team had to think about back when a thousand users was the ceiling instead of the floor. Everyone's remote. And the group chat that used to hold the entire company quietly stopped being able to.

"Going from founder mode to CEO mode has been a challenge," she says. "I was so used to working in the business, doing everything myself and not delegating anything. How do I step aside and actually be a leader, not just a founder?" The stuff that used to live in four people's heads now has to live somewhere the rest of the team can find it: SOPs, standing one-on-ones, Slack updates, meeting notes that are no longer optional. A lot of this year has gone into that kind of unglamorous scaffolding, the systems and documentation that never show up in the product but decide how good it can get.

Natalie will tell you she's type B, which makes the discipline she's building into her days a choice instead of a reflex. She started blocking CEO strategy sessions into her calendar and made them non-negotiable, actual protected time to work on the company instead of inside it. The blocks bought her altitude. Instead of spending the day on customer support and influencer approvals, she started thinking past the current week, about the roadmap and where Rella is actually going. She couldn't count on the habit showing up on its own, so she built it a structure. Same move she made with Ella. Same one Rella makes for the people who use it: take the intentional thing and build something that makes it happen anyway.  

The slow build

Rella was an idea in 2020, a product in January 2022, and an AI company only in 2025. Its four co-founders (Natalie, Natacha Bomparte, Connor Boyce, and Nick Kane) built it with a team where more than half went to NC State, which is not the origin story Silicon Valley usually tells about itself. None of it happened fast. That was kind of the point.

The audience that's followed her for years is still watching, and what they get is a founder in no rush to impress anyone except the people actually using the thing. Which, when you think about it, is the only audience that was ever going to matter. She's building Rella the way she always has. In the open, and only once it's ready.

About HubSpot for Startups

HubSpot for Startups helps early-stage founders build the GTM engine that grows with them, from first customers through Series A and beyond. Eligible startups get access to HubSpot's full platform at a significant discount, plus the tools, content, and community to turn early traction into lasting growth. 

Rella is one of 35,000 startups growing with HubSpot for Startups. Learn more at hubspot.com/startups.

Send DataSend DataSend DataSend DataSend DataSend Data