THE DOOR THAT OPENS THE DOOR
Operational · Private Beta · Provisional Patent Filed

Deterministic post-production
finishing infrastructure.

Raw cut in. Two broadcast-ready masters out — 16:9 and 9:16 — from one fixed pipeline. No colorist. No reference clip. No revision loop. Same input, materially consistent output, every time.

9
Fixed pipeline steps
2
Broadcast masters per run
0
Reference clips required
$28B+
Global post market
Request Investor Brief → Test Out the Pipeline →
The Product

Finishing acceleration.
Not creative replacement.

AutoFinish handles the technical finishing work that happens after an edit is locked — the part that currently requires a colorist, an audio mixer, a version editor, and hours of QC. It doesn't touch the creative cut.

Deterministic Output
Same source file + same settings → materially consistent output class. Not a black box. Every stage logs its decision. Re-runnable.
📡
Broadcast-Legal on First Pass
Luma 16–235 and chroma 16–240 clamped per ITU-R BT.709. Two-pass loudness normalization to −14 LUFS. Ready for air without a QC house.
📐
Dual-Format Delivery
16:9 and 9:16 ProRes 4444 masters from the same graded intermediate — not two separate color sessions. One render, two deliverables.
🎬
Genre-Aware Intelligence
Classifies footage as drama, doc, action, promo, comedy, or news. Applies genre-matched grading parameters. No reference clip required.
🎨
Skin Tone Protection
Skin tones protected across the full ethnic spectrum. ΔE color-science guard with automated reprocess trigger if grade drifts beyond threshold.
.cube LUT Export Every Run
Every render exports a 33×33×33 .cube LUT. Editors can apply it in DaVinci Resolve, Premiere Pro, or any LUT-compatible NLE.
The Pipeline

Nine steps. Fixed order.

Stage order is contractual — reordering is a breaking change requiring patent counsel review. This determinism is a feature, not a constraint.

01
Camera Detection & LUT Normalization
Detects camera model from file metadata. Applies the correct technical input LUT from a 15-camera library (ARRI, Sony, RED, Canon, Blackmagic, DJI, and more). Normalizes to a consistent starting point before any grading begins.
15-Camera Library
02
Ingest & Media Validation
Frame count, FPS, resolution, codec, color space, and audio stream analysis. Every render starts with the facts. Invalid or unsupported media fails explicitly at this stage — no silent downstream errors.
Explicit Validation
03
Source Analysis
Perceptual analysis of source characteristics: exposure, haze, skin detection, contrast range. Results inform grading parameters downstream. Loudnorm pass 1 runs concurrently in background.
Concurrent Loudnorm
04
Color Grade — AI Colorist
Technical pre-normalization (P2/P98 levels stretch, grey-world white balance), followed by genre-aware AI grading: histogram matching, S-curve finish. Vision QA loop validates output against ΔE threshold. Exports a 33×33×33 .cube LUT per render.
AI Colorist · Vision QA
05
Audio Normalization
Two-pass loudnorm to −14 LUFS (EBU R128 / Netflix standard). Pass 1 measurement was pre-computed in Stage 3. Pass 2 applies the calibrated filter. Near-silent streams stream-copied without re-encode.
−14 LUFS · EBU R128
06
Broadcast Legal Clamp
Pixel range enforced to 16–235 (luma) and 16–240 (chroma) per ITU-R BT.601/BT.709. Fused directly into the render pass — no extra encode/decode cycle. Ready for air on first pass.
ITU-R BT.709
07
Smart Reframe / 9:16 Crop
Face detection and saliency-based smart crop for the vertical cut. Same graded intermediate as the 16:9 master — not a separate grade session.
Face Detection
08
Dual Render — ProRes 4444
Parallel render of both formats from the same source. 16:9 master and 9:16 cut. Output codec is configurable: ProRes 4444 (broadcast master), ProRes 422 (delivery), or H.264 (web/social).
Parallel · ProRes 4444
09
QC Validation & Delivery
Before/after comparison frame, job manifest, and stage-level log attached to every render. Files delivered to the members portal. Explicit pass/fail per stage. No silent deliveries.
Stage-Level Logs
Pipeline Contract: Stage order is fixed and non-negotiable. Reordering constitutes a breaking change subject to patent counsel review. This contract is the basis of the reproducibility guarantee — same input, same settings, materially consistent output class.
Return on Investment

Manual finishing costs
$375–$2,100 per deliverable package.

For a standard professional deliverable (one source edit → color, audio, legal clamp, 16:9 master, 9:16 cut), manual finishing requires coordination across multiple roles. AutoFinish collapses that workflow into a single unattended run. The table below uses U.S. market rate ranges as of 2025–2026.

4–13 hrs
Labor hours saved per project
Modeled from market rates — designed to save until production sample data confirms
$325–$2,050
Estimated cost reduction per finished package
Conservative to expected range — excludes revision rounds and coordination overhead
2–5 days → 45 min
Turnaround improvement
Unattended pipeline run — operator time under 30 min for upload + review
Role / Component Manual (Typical) U.S. Rate Range Manual Cost Est. AutoFinish Confidence
Colorist
DaVinci Resolve / Baselight
2–5 hrs $150–$400/hr $300–$2,000 Included in pipeline Modeled
Audio Mixer
Post / production
1–3 hrs $80–$100/hr $80–$300 Included in pipeline Modeled
Editor / AE
Versioning, reframe, QC, export
2–6 hrs $24–$150/hr $50–$900 Included in pipeline Modeled
Software
DaVinci Resolve Studio / Adobe
Ongoing $295 one-time / $20–$35/mo Amortized $29–$299/mo all-in Observed
Total per deliverable package
16:9 master + 9:16 cut · 1 revision round
5–14 hrs $75–$150/hr blended $375–$2,100 <30 min operator
+ ~20 min pipeline
Pilot est.
ROI by buyer segment
Segment Volume Manual cost / mo AutoFinish cost / mo Monthly savings Confidence
Solo editor / creator
4 deliverables/mo · 1 revision
4 projects $1,500–$4,200 $29–$99 $1,400–$4,100 Modeled
Boutique post house
12 deliverables/mo · 2 revisions
12 projects $4,500–$12,600 $99–$299 $4,200–$12,300 Modeled
Enterprise / broadcast facility
40+ deliverables/mo · ongoing
40+ projects $15,000–$50,000+ $299 + API tier (TBD) $14,700–$49,700+ Modeled
Modeled labor savings scenarios
Conservative
5 hrs manual → 3 hrs assisted
~40%
labor reduction
Expected
9 hrs manual → 3.5 hrs assisted
~61%
labor reduction
High-Friction Source
14 hrs manual → 5 hrs assisted
~64%
labor reduction
Modeled from US 2025–2026 market rate ranges and common short-form finishing workflows. Actual outcomes vary by source media, revision cycles, and delivery requirements.
Assumptions & confidence labels: U.S. market rate ranges sourced from 2025–2026 industry benchmarks (BLS, Glassdoor, Production Hub, IATSE rate cards). Flame/Smoke: Smoke is discontinued — current benchmark is Flame / Flame Assist ($600–$900/day, $95–$145/hr observed range). "Observed" = confirmed in production. "Pilot est." = from early-stage runs. "Modeled" = projected from published rate ranges. Actual cost varies by market, revision rounds, and deliverable complexity. Until sufficient production sample data is collected, all savings figures are presented as "designed to save" projections.
The Market

Post-production is a $28.6B market.

Growing to $45B by 2030. Every broadcast deliverable needs color, audio, and legal finishing. Almost none of it is automated end-to-end.

$28.6B
Global Post Market (2025)
Growing to ~$45B by 2030. Every frame produced for broadcast needs finishing.
90%+
Content Consumed Vertically
Every broadcast spot needs a 9:16 cut. Dual-format delivery is now table stakes — almost none is automated.
~$0
Automated E2E Finishing Solutions
No product delivers color + audio + legal clamp + dual-format output in a single deterministic pipeline without operator oversight at each stage.
Competitive Moat

Five things nobody else does.
In one pipeline.

01
Reference-Free Grading
No reference clip required. Ever. Every competing AI color tool requires one — or a human colorist to substitute for one.
02
Genre-Aware Intelligence
Classifies footage type and applies genre-matched grading. No competitor does autonomous genre detection + matched grading in a single pipeline.
03
Full-Spectrum Skin Protection
Calibrated skin tone protection across the full ethnic spectrum with ΔE color-science guard and automated reprocess trigger.
04
Broadcast Legal on First Pass
Luma + chroma clamped per ITU-R BT.709. Ready for air without a QC house. Most AI tools require post-processing for legal delivery.
05
True Dual-Format Output
16:9 + 9:16 ProRes 4444 from the same graded intermediate — not two sessions. One pipeline run, two broadcast masters.
IP Protection
Provisional Patent Filed
March 8, 2026
  • Applicant: Jeffrey Daniel Wright
  • Counsel: Outlier Patent Attorneys
  • Utility filing deadline: March 2027
  • Provisional establishes priority date on the 9-step deterministic pipeline architecture
  • Any architectural change touching pipeline invariants requires Outlier review before implementation
Capability comparison
Capability AutoFinish NLE AI Plugin AI Color Tool AI Video Gen Suite AI
Reference-Free Grading partial partial
Genre-Aware Intelligence
Skin Tone Protection (full spectrum) partial partial
Broadcast Legal (Luma + Chroma)
Dual-Format (16:9 + 9:16) Single Run
Go-to-Market

Three phases.
Production-led growth.

The product sells itself when an editor sees the before/after on their own footage. Phase 1 is working. Phase 2 starts when quality is consistently at the bar.

Live Now
Phase 1 — Curated Private Beta
Hand-selected editors and post houses. Founder-led onboarding. Every job reviewed for quality. Building the quality bar and the testimonial foundation before self-serve opens. Target: 10–20 paying studios at Studio tier ($299/mo) before Phase 2 opens.
Q3 2026
Phase 2 — Self-Serve + Post House Push
Open signup with free Learn tier (watermarked renders) as top-of-funnel. Direct outreach to boutique post houses on Studio tier. Target market: editorial houses doing 10–40 deliverables/month who need consistent quality without a full finishing staff.
Q4 2026
Phase 3 — API + White-Label Partnerships
API access for broadcast facilities and production SaaS platforms that want AutoFinish as an embedded finishing layer. White-label option for post houses that want to offer the pipeline under their own brand. Enterprise pricing TBD by deal size.
Current Status

What's live.
What's next. No overclaiming.

AutoFinish is operational in private beta. The pipeline runs. The portal works. The billing is live. Here is the honest capability state as of May 2026.

Pipeline & Core
9-step pipeline executionLive
Dual-format render (16:9 + 9:16)Live
Audio normalization (−14 LUFS)Live
Broadcast legal clamp (ITU-R BT.709)Live
Genre detection + AI color gradeLive
Vision QA loop with ΔE guardLive
.cube LUT export per renderLive
15-camera technical LUT libraryLive
Platform & Roadmap
Members portal (upload, status, download)Live
Stripe billing (all 4 tiers)Live
Stage-level job logsLive
Before/after comparison frameLive
API access (external)Q3 2026
Batch processingQ3 2026
White-label tierQ4 2026
Multi-user team accountsQ3 2026
Risk & Mitigation

Known risks.
Honest answers.

Risk Mitigation Level
Output quality variance
AI grading results are not uniform across all source types. Some footage categories perform better than others in early testing.
Vision QA loop with per-render ΔE scoring. Automated reprocess trigger when output drifts beyond threshold. Quality gate before delivery. Improving with each production run.
Medium
IP challenge from larger player
Larger companies (Adobe, Blackmagic, Autodesk) could attempt to develop a competing pipeline and challenge the patent.
Provisional filed March 2026 establishing priority date. Utility filing deadline March 2027. 9-step fixed-order pipeline architecture as core claim. Outlier Patent Attorneys engaged.
Medium
NLE integration dependency
Editors prefer to stay inside their NLE (Premiere, Resolve). A standalone upload portal adds friction for embedded workflows.
Pipeline is workflow-agnostic — accepts any export. Phase 3 includes NLE plugin / API layer. .cube LUT export works natively in any color-capable NLE today.
Low
Single-founder execution risk
Early stage is founder-dependent. Key-person risk on product direction and quality standards.
16 years of professional post-production context is the product's moat, not a liability. Building to self-serve model reduces founder-in-the-loop requirement as quality stabilizes.
Medium
Market ceiling concern (niche)
Post-production finishing is a specific workflow step, not a mass consumer market. TAM may be perceived as limited.
$28.6B market with no end-to-end automated solution. API + white-label tier in Phase 3 embeds AutoFinish as infrastructure for other platforms — expanding effective TAM beyond direct users.
Low
Claim risk (early-stage overstatement)
ROI projections and performance claims are modeled from market rate ranges, not a large production sample. Investor or customer conversations could challenge the data.
Explicit confidence labels on every claim: Observed / Pilot estimate / Modeled. Conservative assumptions throughout. "Designed to save" framing until production sample data confirms. Detailed assumptions available in the investor brief.
Low
The Team

Built by the person
who knows the problem.

AutoFinish was not built by software engineers guessing what editors need. It was built by a Senior Editor who ran the finishing workflow for 16 years and decided to automate the parts that shouldn't require a human.

JW
Jeff Wright
Founder · Senior Editor, Fox Entertainment
16 years in post-production at Fox Entertainment. Pattern recognition across thousands of finishing jobs — color, audio, delivery, versioning — is the foundation this pipeline is built on. Every product decision filters through: would this be defensible to a professional finishing house? Provisional patent filed March 2026.

The pipeline is running.
Let's talk.

Whether you're looking at this as an investor, a post house evaluating the Studio tier, or a strategic partner — we're in private beta now and taking conversations.

Request Investor Brief → Test Out the Pipeline →
Direct: [email protected]  ·  (323) 417-0463  ·  autofinish.ai ↗