Here’s a number that caught me off guard. A YouTube channel sitting just under 1 million subscribers is making about $6,000 to $7,000 a month from AdSense. That’s the raw figure the creator behind this AMA, Matt Wolfe, pulled up on screen in his behind-the-scenes video. And it flips the whole fantasy of big YouTube money on its head.
The original poster walked through his dashboard on camera, showing 2024 as his peak year and 2026 so far tracking lower. He’s been intentionally slowing down to one main video per week (his Friday AI news breakdown), which trades view volume for being the single signal-through-noise channel his audience relies on. Brand sponsorships pay more than AdSense, but he kept those numbers private.
The insight breakdown
The creator’s whole operation is built around minimizing post-production so he can ship weekly without burning out. His one-hour recordings collapse into 23-minute final cuts. He uses a Stream Deck to switch scenes live while recording, so most of the edit is already baked in before he stops the camera. The tools doing the heavy lifting: Recut to strip silence from a 65-minute take down to 27 minutes in one click, DaVinci Resolve for the final polish, and a custom zoom preset pack called Greg’s Presets for those on-screen text callouts.
For his viral claw-grab intros, he shared the exact recipe. He records a starting frame (himself off-camera) and an ending frame (himself seated), exports both as stills from DaVinci, then feeds them into Leonardo with a text prompt. He cycles through VO 3.1, Cling 3.0, and Seed Dance 2.0 until one nails the shot. Audio comes baked into the VO and Cling generations, so no separate sound design needed.
Three practical applications you can steal
- 🔹 Batch-test video models instead of committing to one. The expert runs the same prompt through three different generators (VO, Cling, Seed Dance) and keeps the winner. Single-model loyalty means settling for worse output.
- 🔹 Use first-and-last-frame generation for any transition shot. Starting frame plus ending frame plus a short action prompt gets you a usable 5-second clip without needing full scene control. Works for product reveals, talking-head intros, or B-roll inserts.
- 🔹 Build your own news intake stack. The LinkedIn creator uses Feedly for 200-400 daily items across company blogs and AI newsletters, a curated X list called “AI is Awesome,” and Raindrop to tag-and-sort anything worth keeping. A Make.com automation turns saved links into TL;DR summaries on his site.
Tips and pitfalls
On the AI-creator panic: the one who posted it doesn’t buy the doom narrative. He argues that when a video feels obviously AI-voiced or AI-visualized, he clicks off within seconds, and he’s the AI guy. His bet is that human commentary stays valuable even as slop proliferates, because Gen Alpha’s taste will mature the same way every previous generation’s did.
On the “faceless AI YouTube channel” pitch floating around: this industry pro pushes back hard. He says the creators actually winning with AI-generated shorts are running real storytelling experiments, not just cranking prompts through a pipeline. Most of the get-rich automation pitches miss the part where you still need a hook, a narrative arc, and iteration on what landed.
On ChatGPT vs Anthropic: the person who shared it doesn’t think OpenAI is bleeding. ChatGPT is still the Kleenex of AI for normal users. What Anthropic did smart was pour everything into better coding models, because coding ability unlocks tool use, memory, and agent behavior downstream. He thinks OpenAI just caught up to that insight, which is why Sora and their other experiments got deprioritized.
On staying current: his answer was refreshingly simple. Watch one weekly roundup video and skip the daily grind. If you want the full firehose, subscribe to Feedly with every major AI company blog plus the Verge, TechCrunch, The Neuron, The Rundown, and AI Breakfast.
Gear and stack the creator revealed
Desktop: Mac M3 Ultra for main work, PC with RTX 5090 for Windows testing, DGX Spark with 128GB VRAM for local models. Travel: MacBook M4 Pro. For app building: Cursor for custom scripts (he vibe-coded the AMA overlay tool used in this very video), N8N for the Future Tools website automation, Make.com for news database updates. Voice dubbing into other languages is coming soon; he’s in talks with ElevenLabs and Ditto.
The takeaway worth keeping
The savvy professional behind this channel spends far more time recording than editing, runs three video models in parallel for every intro, and treats one well-made weekly video as more valuable than five rushed ones. The AdSense number is the headline, but the real lesson is that a small focused workflow beats volume for almost everyone reading this.
Check the full video for the live demo of his editing pipeline, the studio walkthrough, and the exact automations running his site.