Models crossed the bar
Veo, Kling, Seedance, Runway, and Higgsfield now produce single shots good enough for broadcast. The clip is solved. The film is not.
THE PRODUCTION LAYER FOR AI VIDEO
Your vision.
Every frame.
From the first spark to the final cut.
Locked characters. Unbroken continuity.
Any model.
SCROLL TO DISCOVER
+CREATIVE FREEDOM.
PRODUCTION PRECISION.
PRODUCTION PRECISION.
ONE PIPELINE. ANY MODEL.
HiggsfieldKlingSeedanceAnd whatever comes next
01 / THE VISION
Great clips are
just the beginning.
A face changes. A jacket vanishes. The light resets. One beautiful clip becomes fifty disconnected shots.
Your story deserves to hold together.
AI CONCEPT GALLERY01 / 03
After hours.City light. Human stories.
Fashion filmAI concept gallery · Illustrative campaign directions.
The hard part isn’t shot one.
It’s shot forty still matching.
Lockframe carries your characters, world, and creative decisions through the whole production. The details stay locked. The possibilities stay open.
See how continuity holds02 / THE PIPELINE
You direct.
Lockframe connects.
Seven stages. Two creative approvals.
One continuous production.
HUMAN GATE 01
Every great film starts with a story.
Turn the idea into a structured narrative and shot list. Shape the story, approve the script, and give every shot a purpose.
OUTPUTStory → structured shot list
NARRATIVE / SHOT LIST
ILLUSTRATIVE PRODUCTIONLF—00101Establish the worldWide
02Introduce the characterMedium
03Move the story forwardTracking
You approve the script and shot list.
Human approvalAutomated pipeline
The creative decisions stay yours.
03 / BUILT TO HOLD TOGETHER
Continuity.
By design.
Consistency isn’t another line in a prompt. It’s the foundation of the entire production.
ONE WORLD. EVERY SHOT.CREATIVE LOCK
Illustrative asset registry
01RegistryA shared memory for your world.
Every character, prop, location, clip, and grade is referenced only by @tag. Prose descriptions inside prompts are lint errors.
@sarah_chen · @jacket_worn · @loft_dusk02DeterminismThe same character. Word for word.
Character locks are pasted verbatim by string substitution. No model ever "rephrases for context." Paraphrase is where drift enters.
substitute(@sarah_chen) → verbatim03PortabilityYour vision travels between models.
Shot records store intent (2.39:1, structured audio spec), never vendor dialect. Switching models is an adapter, not a rewrite.
ir { ratio: "2.39:1", audio: {...} }04VerificationEvery shot earns its place.
Every compiled prompt passes deterministic lint plus an adversarial critic in a fresh context. The author model never reviews its own work.
lint() → critic(fresh_ctx) → submit04 / MADE FOR YOUR NEXT CAMPAIGN
For teams with
more to make.
Make the film you imagined.
Then make it work everywhere.
AI-generated campaign concepts.
THE PRODUCTION BOTTLENECK
- Shoot or agency build: weeks per spot
- Five- to six-figure production per campaign
- Every variant, market, and format is a new job
- AI experiments: one clip at a time, re-rolled by hand
THE LOCKFRAME APPROACH
- Days from approved script to finished cuts
- A fraction of the production line; cost preflight before every job
- Locked characters and worlds reused across every variant
- Machine QC catches the drift before anyone sees it
05 / THE BIGGER PICTURE
The models make clips.
We’re building what’s next.
Model vendors are racing on clip quality. The opportunity is the production layer that brings the whole film together.
RAISING PRE-SEED
$2MUSD
A working seven-stage pipeline.
18 months to the next chapter.
More video, same budget
Brands and agencies are asked for more variants, more formats, more markets every quarter. Production cost is the constraint.
The best model changes monthly
Betting a workflow on one vendor is a liability. Teams want a layer that survives the next model release.
01The landscape
Where Lockframe fits
Where we sit
Everyone else optimizes a clip. We orchestrate the film.
| Capability | Model vendors | Creative / edit tools | AI studios | Lockframe |
|---|---|---|---|---|
| Single-clip quality | Yes | Partial | Yes | Yes |
| Multi-shot continuity | Partial | No | Partial | Yes |
| Model-agnostic | No | Yes | Partial | Yes |
| Automated consistency QC | No | No | No | Yes |
| Rules that compound across productions | No | No | Partial | Yes |
When a vendor ships native multi-shot:
They improve one model. We orchestrate all of them, keep the customer’s characters and rules portable, and adopt the new capability through an adapter.
02The business
Pilots, platform, and usage
How we make money
Pilots fund the roadmap. Platform plus usage is the recurring engine.
Indicative pricing. Manual steps in pilots become the product backlog.
Platform
$2–5k
/ month per team
Subscription per team or per production. The registry, lint, critic, QC, and assembly layer.
Generation usage
+20–30%
on provider spend
Provider spend flows through the pipeline with cost preflight on every job. We take margin on the volume we orchestrate.
Pilots
$15–40k
per pilot
Forward-deployed: we run the first campaign with the customer. Paid, scoped, and designed to convert to platform.
$1T+
global advertising spend in 2026, the first time past a trillion
Dentsu Global Ad Spend Forecasts, Dec 2025
$82B
US digital video advertising spend forecast for 2026, up 11% year over year
IAB, 2026
Where we start
Every dollar of video media spend sits on top of production spend that the market treats as a cost center. Our wedge is the production line for branded video at agencies and in-house studios, where volume is rising and per-spot budgets are not.
Long-form, game cinematics, and interactive education follow on the same pipeline.
03The roadmap
Built today. What comes next.
Where we are
The pipeline is built and running end to end.
Built
- Seven-stage pipeline, logline to assembly, behind two human gates
- Tagged, versioned asset registry with type-checked references
- Deterministic lint plus adversarial critic pass on every prompt
- Provider-neutral shot IR with adapters for Higgsfield, Kling, and Seedance
- Consistency QC with retry, asset repair, and a human queue for exhausted shots
- Failure log that promotes recurring diagnoses to lint rules
- Cost preflight on every job, with usage reporting per production
Next 90 days
- 01First branded campaign delivered end to end, published as the reference case
- 023–5 paid pilots with agencies and in-house brand studios
- 03Run deterministic detectors (face embedding, optical flow) on hosted outputs
- 04Rate cards calibrated against provider invoices
- 05Pilot learnings turned into the v1 platform backlog
Roadmap
Same pipeline, same registry, same rules.
Each phase is a new renderer and a new buyer, not a new product.
Q3–Q4 2026
Campaign video
Brand and agency pilots. Locked characters and worlds across every variant, format, and market.
2027
Long-form and games
Shorts, series, and episodic content. Cinematics and narrative pipelines for game studios, same registry, different renderer.
2028
Interactive learning
Character-driven interactive content for learning. Open shot IR spec and consistency benchmark published under a foundation.
04The round
Team and seed milestones
The ask
$2M
pre-seed
18 months of runway to reach seed milestones.
Milestones to seed
- 0110+ paying teams on the platform
- 02Repeat campaigns from the first pilot customers
- 03Deterministic QC running on every shot, no manual review past Gate 2
- 04Adapters for every major video model, added in days, not weeks
- 05Lint rule library measurably cutting retries per shot
- 06Open shot IR spec and consistency benchmark published
Team
Oleksii Vergunenko
Founder and CEO
Software engineer with a physics background and a creative practice. Builds systems from first principles.
First hires with this round
Head of production
Runs the pilot campaigns and the studio arm
Head of sales
Agency and brand relationships, pilot-to-platform conversion
YOUR NEXT CHAPTER
Make something
worth watching.
Lock your characters. Follow your vision.
Ship the whole film.
AGENCIES. BRANDS. BOLD IDEAS.
ov@generationix.net