inference academy ← inference.kg
ENRU
Apply — Cohort 01
inference academy — track 01 Cohort 01 — September 2026 Live · RU + EN

AI
Creator

Direct image, video, music and voice models like a one-person studio. Taste required — design degree not. Learn to make media with intent and consistency. Four weeks later, you launch a channel or campaign built entirely by you.

Apply — 20 seats → Read the program ↓
6
Weeks, live cohort
6
Artifacts you’ll ship
0
Design background required
3×
Live sessions weekly
1
Channel launched at demo day
01Outcomes first

What you’ll ship.

Not “topics covered” — things made. Every week ends with a published piece with your name on it. By demo day you have a six-piece portfolio and an engine that keeps producing.

Week 1

A visual series

Twelve images in one deliberate, consistent style — yours. Subject, composition, light and mood under your control, published as a set.

Week 2

A brand kit

Name, logo, palette, product shots and ad creatives for a real or invented brand — coherent enough to use tomorrow.

Week 3

A brand film

A 30–60 second film: storyboarded, generated shot by shot, and cut by you. Camera language included.

Week 4

A voiced & scored video

Your film with voiceover, music and sound design — finished in Russian and English. Yes, both.

Week 5

A content engine

A 30-day content calendar plus ten assets, batch-produced in one sitting with templates and pipelines you built.

Week 6

Your channel, live

A channel or campaign launched at demo day — first posts up, engine running, numbers honest.

02Who it’s for

Made for people with taste
and no production team.

The bottleneck in media was never ideas — it was production. The models removed it. What’s left is what this course teaches: taste, direction, and consistency.

You’ll fit right in

  • ✓Marketers & SMM — make a month of content in a week, in-house.
  • ✓Designers & illustrators — add motion, sound and speed to taste you already have.
  • ✓Founders & small business — brand and content without an agency retainer.
  • ✓Creators & bloggers — scale output without losing your voice.
  • ✓Total beginners with taste — the tools finally meet you halfway.

Prerequisites

  • 01A laptop from the last ~5 years. The models run in the cloud.
  • 026–8 honest hours a week, for four weeks.
  • 03Zero design or Photoshop experience. We start from “what makes an image good”.

Not sure this is your track? The Agentic Engineering track covers software — websites, apps, games. Same method, different craft. Or just ask us on Telegram.

03The program

The program, mapped.

Each module: two live studio sessions, your build days, and a Saturday review. Open a week to see exactly what you’ll learn and make.

Week 01Seeing like a model+

What you’ll learn

  • How image models think — and why “beautiful photo” is a useless prompt.
  • Prompt anatomy: subject, composition, light, lens, mood, style.
  • Reference boards and taste: stealing like an artist, ethically.
  • Consistency: keeping a character, product or style stable across a series.
You ship

A 12-image visual series in one deliberate style — your first portfolio piece, published.

Week 02Identity+

What you’ll learn

  • Brand systems with AI: name, logo, type, palette — why identities feel coherent.
  • Product photography without a camera: angles, surfaces, light.
  • Ad creatives that pass the squint test: hierarchy, contrast, one message.
  • The editing pipeline: inpainting, outpainting, upscaling, cleanup.
You ship

A complete brand kit for a real or invented brand — from logo to ad set.

Week 03Motion+

What you’ll learn

  • Image-to-video: motion prompts and camera language — push, orbit, rack focus.
  • Storyboards: planning shots before spending generations.
  • Continuity: keeping characters and worlds stable between shots.
  • Editing rhythm: cuts, pacing, and music as structure.
You ship

A 30–60s brand film — storyboarded, generated, cut by you.

Week 04Sound+

What you’ll learn

  • Voiceover that doesn’t sound like a robot: direction, emphasis, pacing.
  • Music generation: from brief to track — and when to license instead.
  • Sound design: the last 20% that makes video feel finished.
  • Multilingual content: dubbing yourself into Russian, English and beyond.
You ship

Your film, fully voiced and scored — delivered in two languages.

Week 05The Engine+

What you’ll learn

  • Content systems: from one idea to a 30-day calendar.
  • Batch production: templates and pipelines that make ten assets in a sitting.
  • Platform grammar: Reels, Shorts, TikTok, Telegram — same idea, native form.
  • Pricing your work: client packages, scope, and saying no.
You ship

A 30-day content engine — calendar plus ten assets, batched.

Week 06Capstone — the launch+

What you’ll learn

  • Positioning a channel: who it’s for and why they’d follow.
  • The first ten posts: launching without an audience.
  • Reading the numbers kindly: what to change, what to keep.
  • Demo day prep: presenting creative work so it lands.
You ship

Your channel or campaign — live, with the engine behind it. Presented at demo day.

04Price & guarantee

One price.
Nothing hidden.

Cohort 01 — founding price
$500
≈ 43 500 сом · two installments at no extra cost
The honest guarantee: attend the sessions, attempt the work — and if by the end of week 2 you decide it’s not for you, we refund everything. No forms, no hard feelings.

What’s included

  • ✓18 live events: 12 studio sessions + 6 review days, evenings Bishkek time.
  • ✓Personal review of every weekly artifact by the lab.
  • ✓The lab’s prompt library — 1,004 curated, working prompts, and our reference boards.
  • ✓Cohort chat with the lab on call; recordings of everything.
  • ✓Demo day, the academy showcase, certificate, alumni community — and a path to create with the lab.
Tools: plan on ~$30–60/mo of AI subscriptions during the course. Free routes shown where they exist.
05Course FAQ

Asked & answered.

Q1I can’t draw and I’ve never opened Photoshop.+
Perfect — there’s nothing to unlearn. The course starts from “what makes an image good”, builds your reference habit, and trains taste through weekly critique. Drawing skill was the old gate; direction is the new craft.
Q2Which tools exactly?+
The current best per medium: image models of the Midjourney / GPT Image class, video of the Veo / Sora class, voice and music of the ElevenLabs / Suno class. We’re tool-honest: models change every season — the craft we grade (brief, reference, critique, consistency) transfers to whatever ships next.
Q3Is AI content even allowed? Will platforms bury it?+
Platforms reward what people finish watching — that’s the bar we train for. We cover disclosure norms and each platform’s current policy, and we skip the tricks: the durable advantage is making things people actually want to share.
Q4Can I sell this as a service afterwards?+
Yes — week 5 covers packages, scope and pricing, and several capstones are built as client work from day one. A month of content for a local brand is a real, sellable product in this market.
Q5What about rights — faces, voices, copyright?+
Covered explicitly: what each tool licenses you to use commercially, consent for voices and likenesses, and the lines you don’t cross. You’ll leave knowing what you own and what you must clear.

Your studio opens
this autumn.

Cohort 01 — September 2026, 20 seats, reviewed in order of application. Tell us what you dream of making; that’s the whole application.

Apply via Telegram → Apply by email
$500 · installments available · honest week-2 refund