AI technology company

We build AI infrastructure for video content production — and deploy it across 5 live YouTube channels.

Proprietary pipeline: research → script → images → voice → assembly → publish. Script-to-publish in one working day. Built on Google Cloud.

See pipeline output How we build it

active YouTube channels

videos published per week

~9min

avg episode length, script-to-publish in 1 day

months of daily pipeline iteration

Pipeline Output

What the pipeline actually produces.

Research → script → 122 AI frames → voice clone → assembled. Each episode took one working day. Same infrastructure, different channel.

Frugal Forward

Her $42K Paycheck Cost Us $4,000. Two-Income Trap.

Frank LoRA 122 frames FLUX-fp8 5:48

The Lazy Grocer

Dollar Tree's NEW $1.50 Garden Products — What Actually Works

Maggie LoRA 122 frames FLUX-fp8 8:31

Inner Mechanics

Ferrari Brain. ADHD.

Dave LoRA 122 frames FLUX-fp8 5:34

Lived

Rasputin — The Night He Wouldn't Die

Watercolor stack 122 frames LTX-Video I2V 6:23

Lived

You Are on Dyatlov Pass: From Nightfall to Dawn

Watercolor stack 122 frames Same pipeline 6:03

Channels

Five live deployments. One pipeline.

Each channel is a live deployment of the same technology stack — different persona, different niche, same infrastructure. Finance, gardening, psychology, history. One pipeline adapts to all.

Frugal Forward

Personal finance

Persona: Frank LoRA
Visual: Hand-drawn cartoon
Status: Live · weekly

FF ↗

The Lazy Grocer

Practical gardening

Persona: Maggie LoRA
Visual: Painterly photoreal
Status: Live · weekly

LG ↗

Inner Mechanics

Psychology · PMC-cited

Persona: Dave LoRA
Visual: Minimalist cartoon
Status: Live · weekly

MH ↗

Lived

Historical POV · 2nd-person

Persona: Voice clone (narrator)
Visual: Watercolor · Rackham/Turner
Status: Live · weekly

YA ↗

Rhymes of History

Historical pattern-recognition

Persona: Chronos
Visual: In development
Status: Pilot · validation

RH Pilot

Economics

Digital-native business model. Per-channel math.

Revenue scales with audience — no physical production overhead, no studio costs. Each channel is an independent revenue unit.

Primary

YouTube ad monetization

Pays per 1,000 views (RPM). Niche-dependent: finance $5–15, history $3–6, psychology $2–5, gardening $3–7. Scales linearly with views.

Secondary

Brand integrations

Native sponsorships per video, negotiated directly. Rates scale with subscriber count: $1k+ per integration at 100k subs, $5k+ at 500k subs.

Future

Pipeline licensing (2027)

License access to the production pipeline for independent creators. Tiered SaaS model — recurring revenue independent of channel performance.

Per-channel projection

Path to profitability — one channel

Each new channel runs on the same pipeline — production cost stays low while revenue compounds with audience.

Stage 1

Validation

0 – 10k

subscribers · months 1–4

Audience build, format iteration. Minimal ad revenue.

Stage 2

Break-even

10k – 50k

subscribers · months 5–10

Ad revenue covers production and infrastructure. Channel self-sustains.

Stage 3

Profitability

50k+

subscribers · months 10+

Ad rev + brand deals exceed costs. Margin funds next channel launch.

With 5 channels live today and 10+ planned by end of 2026, the model compounds: each new channel adds revenue without proportional cost.

Technology

How one 10-minute episode actually gets made.

Fourteen distinct models and tools, coordinated by ~6,000 lines of orchestration code. AI-native means every step is a purpose-built model — not a wrapper around a single chatbot.

Research

Structured interview protocol over peer-sourced citations (PMC, USDA, NIMH). Source-backed research, not generated facts.

Script

Channel-specific doctrine files enforce voice and narrative structure. Multi-pass quality loop scores tension before sign-off.

Visual prompts

Per-channel visual grammar bakes in palette, composition, anti-AI keywords. Fully automated — no manual prompt editing.

Images + character LoRA

Custom-trained character LoRAs (one per persona) on FLUX base. Identity holds across all frames via conditioning — not per-shot face-swap.

Motion hook

Image-to-video stitch on the opening segment — retains audience past the 8-second drop-off. Speed-matched to the voiceover beat.

Voice clone

A 30-60s reference per channel produces a stable voice clone. Per-segment generation — granular retry, not whole-episode re-rolls.

Assembly + publish

Automated video assembly with SFX, beat-aligned cuts, vertical 9:16 short, and thumbnail variants — all generated from the same source.

Why this is genuinely difficult

Character consistency across 100+ frames without per-shot retouching — solved by training four LoRAs in-house (~30 min/persona on A100), not by buying face-swap APIs.
Long-form narrative coherence at scale — playbook-driven script generation: channel doctrine + reverse-countdown templates + multi-pass quality loop that scores tension before sign-off.
Per-channel aesthetic without re-engineering — config-driven visual stack. One doctrine file = one channel personality. Adding the sixth channel is a config change, not a code change.
Low marginal cost per 10-min episode — aggressive Spot GPU usage, open-weight FLUX/LTX/Qwen models, minimal closed-API spend. Pipeline is engineered for economic viability at scale, not dependent on expensive closed-API calls.

Infrastructure

Built on Google Cloud

Every image, every video frame, every voice clip is rendered on Google Cloud GPUs. Compute Engine with L4 for fast iteration, Spot A100 for batch production. NotebookLM for source-backed research. The entire pipeline lives in one Google Cloud project.

Compute Engine — 2× A100 80GB Spot VMs at production cadence, ~400 GPU-hours/month
Cloud Storage — ~500 GB media archive, version-controlled, lifecycle-tiered
NotebookLM — source-backed research, 50+ citations per topic
Vertex AI — pipeline orchestration, batch jobs, model serving for in-house LoRAs
IAM + Quotas — fine-grained cost and resource control across the project

Target monthly burn at production cadence

~$500–600 / mo

GPU compute (~$480) + storage (~$10) + Vertex AI orchestration. Credits would cover ~18–24 months of operations at the target cadence — directly funding the scaling from 5 to 10+ channels.

Roadmap

From 5 channels to 10+ by end of 2026.

Q2 2026

Now

5 channels active. Pipeline stabilized on Spot A100. 4 videos/week cadence.

Q3 2026

LoRA character consistency improvements. +2 channels in new niches. Multi-language voice synthesis.

Q4 2026

Goal

10+ channels. EN/RU localization. Multi-platform distribution.

2027

Vision

License pipeline to creators. Open-source non-proprietary parts.

Founder

One operator. One vision. Built in the open.

Valerii Serko

Founder & Engineer

Engineering background. Founded S.Ler Group in 2026 to build AI-native content infrastructure — production tooling that turns peer-reviewed research into watchable, sourced video at consumer-facing scale. Based in Almaty. The pipeline you see on this site is the product of 8+ months of daily iteration on every stage from research protocol to final assembly.

LinkedIn Email

Get in touch

Partnerships, credits, investment.

Building something specific? Reach out directly.

valerii.serko@slergroup.com

Valerii Serko, founder

Headquarters

S.Ler Group
Vasilia Klochkova 105
Almaty, 050057
Kazakhstan