Bryan Ng built a YouTube AI platform on the premise that would have sounded wrong three years ago: the subscriber count of a channel barely predicts whether it earns money.

His path to that conclusion started with two years in the Singapore military, wound through a Twitter growth experiment, and ran through a Done-For-You YouTube scriptwriting agency serving more than 100 business owners. Along the way, he identified something more specific than a creativity problem. Most professionals and businesses that wanted to build on YouTube had ideas. What they lacked was a system to carry those ideas all the way through to a published video.

“YouTube is one of those things where it’s so much more difficult compared to short form,” Bryan says. “You need to make sure your idea, your title, your thumbnail, your script, your hook, your recording and your editing are all nailed down.”

Subscribr, which Bryan co-founded in 2025, is based in New Orleans. The platform serves faceless creators, on-camera YouTubers, and businesses looking to establish organic content operations, handling idea generation, scripting, thumbnail creation, and full video production, including an AI avatar option for creators who do not want to appear on camera. Bryan, who serves as CEO, began building his own YouTube presence in mid-2024 after growing an X/Twitter following of 10,000 in six months, and later ran a scriptwriting agency before pivoting to software when AI writing capabilities improved enough to put the agency model under pressure.

Subscriber Count Is YouTube’s Most Obsolete Metric

Bryan’s argument about subscriber counts is not a platform critique. It is an observation about how YouTube’s recommendation engine now works.

“Right now, we’re in the attention economy,” he says. “All attention is always shifting to different ideas, different people. Subscriber count doesn’t really matter too much.”

Five to ten years ago, viewers navigated YouTube primarily through their subscriptions tab, following specific creators closely. Today, the algorithm routes content based on engagement signals, not subscription relationships. A new channel with sharp packaging and a well-constructed hook can outperform one that has spent years accumulating a subscriber base, because the system does not advantage existing relationships over relevant content.

What Bryan calls packaging, the title and thumbnail combination, determines whether a video gets a click from the right audience. According to him, a click from the wrong audience type can suppress a video’s distribution, since YouTube interprets poor audience-content fit as a negative signal. The first 30 seconds carry a different weight on YouTube than on short-form platforms: they must convey the full value proposition of the video and hold viewers long enough to make the content worth recommending.

“Everything always starts with an idea and the packaging,” Bryan says. “Then everything else follows.”

What Bryan Couldn’t Find on YouTube Became the Business

The scriptwriting side of Subscribr traces directly to a knowledge deficit Bryan encountered while still in the military. He attended a networking event in London, connected with a YouTube agency owner, and was hired to learn YouTube scriptwriting. The problem: almost no usable instructions existed.

“When I had gone and searched ‘How to write a YouTube script,’ all the content was fairly basic,” he says.

Bryan synthesized material from copywriting literature, brand strategy, and YouTube-specific sources to develop his own frameworks, then posted that content to his own channel because the instruction was not available elsewhere. His top tutorial on the subject has accumulated 300,000 views.

The agency came from that audience. By the time he was serving more than 100 business owners, he was also watching AI writing tools improve sharply enough to raise real questions about the agency model’s durability. The pivot to software followed. “I really want to get ahead of it,” he says, describing his calculation as the models improved toward creative writing tasks.

The second bottleneck he had run into personally was post-production. His first YouTube video took five hours to produce, a ten-minute piece derailed by camera setup and a microphone that stopped recording mid-session. Professional editing software had no place in his workflow. “That’s a big bottleneck,” he says.

Idea to Upload, Inside One Platform

When a user opens Subscribr, the first step is connecting an existing YouTube channel. The platform builds a brand voice profile from the channel’s history, identifying which topics have performed well and what stylistic tendencies the creator has established. For new channels without a publishing history, users can input a reference creator. Bryan uses the example of inputting Alex Hormozi’s channel as a style reference, after which the platform generates ideas for the user’s niche in a similar format.

Idea generation draws from tracking data across millions of channels, surfacing what is currently performing well by vertical. From a selected idea, the platform’s scripting agent produces a full script without requiring the user to manage individual prompts. Bryan says the system generates 4,000- to 8,000-word scripts in a single workflow, incorporating research, verification, and what he calls a humanization pass.

Thumbnail generation, AI avatar video production, and automated editing of face-on recordings complete the production stack. The commercial case Bryan makes to businesses already spending on paid distribution: Subscribr provides an organic content arm that runs without requiring on-camera talent, a dedicated editor, or production overhead.

Why Expertise Still Matters When Anyone Can Publish

The obvious question about a platform that removes most YouTube production friction is what happens to trust when volume becomes inexpensive to produce.

Bryan’s answer draws a direct distinction between the tool and the person using it. “If you give a veteran who knows what he’s doing a sniper rifle, he’s going to be a lot more effective,” he says, contrasting expert use with distributing the same capability indiscriminately.

Authority, in his framework, is the standing that makes a creator’s output worth watching in the first place. Credentials, track record, and visible expertise signal to viewers that the content merits their time. Producing polished scripts for someone without standing in a field does not create that standing.

The second variable is what Bryan calls unique insights: experiences, perspectives, or learnings that are genuinely proprietary and cannot be reconstructed from publicly available data. A client conversation, a counterintuitive read on a developing story, a personal history that informs a position. “Storytelling has been around for so long,” he says. “If it’s a personal story, that’s how the viewer creates a connection with you.”

For brands evaluating AI-assisted channels as potential partners, he says the operative variable is audience quality, not production method. A targeted, engaged audience justifies investment regardless of how the content was made. What warrants caution is high view counts attached to low-quality or poorly matched audiences.

The Two Shifts Most Creators Have Not Fully Priced In

Bryan identifies two technology waves he sees as underestimated in the Creator Economy.

The first is AI avatars. He says the technology has largely been solved for short-form content and is closing the distance on long-form, pointing to examples he has observed of channels where a human creator has been replaced by a full AI avatar presenter. He attributes rising adoption partly to the basic reality that camera confidence is unevenly distributed. “Not everyone is fluent on camera,” Bryan says. “It’s just not their personality type.”

The second wave is AI video editing. He describes an emerging toolkit capable of taking raw footage and a style reference video and producing a finished, edited output with motion graphics, captions, and transitions, without manual intervention. What accelerates both waves, in his view, is cost. Chinese language models, specifically DeepSeek and GLM-5.2, are cheap enough to make scale testing economically viable in ways that the major U.S. AI providers currently do not allow.

Subscribr is positioning to sit at the center of that production stack as both technologies mature. The goal is a platform where the distance from idea to published video collapses into a single session, regardless of a user’s technical background or comfort with cameras.

“I can immediately go to Subscribr, pay a couple of dollars, and create my very first video and publish it,” Bryan says. For a professional with expertise worth sharing but no appetite for the production side of YouTube, that access, he argues, is what the platform was always built for.

Subscribe to Our Newsletter

Check Out Our Podcast