I am not an agency and not an outsourced contractor. I built my own network of channels from zero, without a single dollar of ad budget, and I know these platforms from the side of someone who grows on them.
Istorium launched on 2 September 2025 on pure storytelling, with no advertising at all. Below is where it stands today. Open it and see for yourself.
39,162,622 lifetime views across 113 videos, averaging 349,679 per video.
TikTok44,000followers1,500,000 likes. The platform does not publish total view counts.
Instagram30,500followers118 posts, verified account.
Plus the English-language Lore Vault channel, same method and same production pipeline.
Accredited by Higgsfield Academy in AI filmmaking. I run the whole cycle myself: script, visuals, characters, voice, music, edit and publishing. That means there is no vendor sitting between the idea and the finished video.
✓Higgsfield Academy · AI Filmmaking
Below are the real working materials from an Istorium episode about the invention of the speedometer. Not presentation slides, but the parts a video is assembled from before a single frame is shot.











Content that does not die within two days
Storytelling built on facts and history. My channel shows it: ten videos still bring in more than half of all traffic months after release.
Volume without losing quality
My own production pipeline. Every unit passes review while the volume keeps growing. That is solved by a system, not by more hands.
A recognisable style instead of a mixed bag
Recurring characters and one visual code. Characters get attached to the brand and build recognition, faces and objects do not drift between videos.
Clear reporting instead of promises
Weekly interactive reports: week-over-week dynamics, referral funnel, platform comparison, progress on every unit of content.
I work with every major generative model used in content production, and I write the automation that ties them together myself.
Claude, GPT, Grok, Gemini
Seedance 2 and 2.5, Veo 3, Kling 3, Gemini Omni Flash, Runway, Hailuo
GPT Image 2, Nano Banana 2, Seedream 5.0, Midjourney, Flux, SDXL, ComfyUI, ControlNet, LoRA
Voice synthesis and cloning, lip-sync, music, sound design, Suno, Mubert
Claude Code, Claude Agent SDK, MCP, Python, FastAPI, JavaScript, REST, Git
Cloudflare Pages and Workers, D1, R2, static sites, SEO and GEO
My own AI content production studio. Designed, coded and shipped to production single-handed.
Product description. Copy provided by Stas.
A live capture of one of the production modes.

Second product mode.

Third product mode.
The same method runs on my own channels and on client work. Below is its skeleton, with no client attached.
The working horizon and what counts as a result at each stage.
Month-by-month breakdown: immersion, ramp-up, scale.
The four weeks of the first month, step by step.
Characters, props, locations: the library every video is assembled from.
How it gains force: from the first videos to accumulated reach.
Red lines, release discipline, areas of responsibility.
Seven directions. Each one holds real delivered work, not examples borrowed from the internet.
A talking avatar speaking for the brand: face, voice, lip-sync, product and motion generated end to end.

Thirteen ad clips with talking avatars: serious, comedic and intro formats.
Live showcase: avatars.cantry.ai
One script performed by four different characters, including a new character built from scratch.
The client raised the rate and asked for more
Vertical clips in testimonial and unboxing formats.

Room for one more piece in this category.
Cinematic short form: light, framing, editing rhythm, original score and voice.

A short film: 7 scenes, cloned voice, original score, 4K.
87 seconds, delivered in one day
A 45 to 90 second reel, best shot first.

Room for one more piece in this category.
Work for game studios: ad creative, characters, interface concepts.

Awaiting a decision on which game work goes public.

Room for a piece in this category.

Room for a piece in this category.
The animation track: one visual code, recurring characters, a serialised format.
An animated series: vertical and horizontal cuts, 4K.

Room for a piece in this category.

Room for a piece in this category.
An account taken apart by data: what gets produced, what gets watched, where money and attention leak.

An anonymised full audit: production versus views, review mining, audience segments.
Live example: otchet.cantry.ai
Room for a piece in this category.

Room for a piece in this category.
Not a one-off video but a system: a pipeline that ships volume and holds quality.

Launched 2 September 2025 on storytelling, with no advertising.
39,000,000+ views, 42,400+ subscribersA capture of the pipeline: from source material to a finished unit of content.

Room for a piece in this category.
Recruiters and buyers increasingly ask a model instead of a search engine. I check and tune what the model actually sees.

Static delivery, Person markup, answer-first FAQ, robots tuned for AI bots, sitemap, llms.txt, two language trees.
Readable by people and by models alike
Room for a client GEO case.

Room for a piece in this category.
Not "everything is fine" but numbers: what shipped, what worked, what gets cut, what happens next week.
Materials provided by Stas: the weekly report format, its sections, examples.
Open to employment and to client projects. Remote, working with teams in the US, UK and Europe.