AI-Powered DaVinci Resolve · Windows

Stop Managing
Footage.
Start Editing Stories.

DocuFlow automates every logistics step of documentary post-production — footage organisation, transcription, soundbite identification, story development, and rough cut assembly — entirely inside DaVinci Resolve. No cloud uploads. No new software to learn.

Turns a 2–3 week pre-edit workflow into a single day.

✓ Footage never leaves your machine ✓ Works with footage you already have ✓ ~$1–5 per full pipeline run
DocuFlow interface showing the AI-powered documentary editing pipeline
96%
faster footage classification
200+
clips organised in under 5 minutes
15s
to find any shot in plain English
1day
from raw footage to rough cut
The Problem

80% of editing time isn't editing

In a standard documentary workflow, an editor spends the vast majority of their time on logistics — before a single creative decision is made.

📂

Day 1–2: Classify & rename footage

200 clips, inconsistent camera names, no bin structure. Manual logging takes 2–4 hours before Resolve is even opened.

2–4 hours
📝

Day 2–5: Log transcripts

Every interview must be transcribed and logged before you can identify soundbites or begin story development.

8–20 hours
🧩

Day 5–10: Story development

Scrubbing through hours of footage, building selects, finding the angle. The creative part, but buried under logistics.

1–3 days
✂️

Day 10–14: Build the rough cut

Finally placing clips on a timeline. By this point, half the budget may already be spent on pre-edit admin.

2–4 hours
How It Works

From raw footage to rough cut in one day

DocuFlow bridges your footage folder and DaVinci Resolve through a structured pipeline. Python handles the heavy lifting; Lua scripts connect everything to Resolve.

01
Python · CLI

Ingest & Organise

Point DocuFlow at your footage folder. It reads every filename, classifies each clip (interview, b-roll, location, people), renames them to a consistent convention, copies them to your project folder, and creates a structured intake_log.json ready for Resolve.

✓ 200 clips classified and renamed in ~5 minutes
02
DaVinci Resolve · Lua

Import & Transcribe in Resolve

Run the DocuFlow import script inside Resolve. Clips land in the correct bins automatically. Then run the transcription script — Resolve's built-in AI transcribes every clip and DocuFlow exports per-clip transcript JSON to your catalog.

✓ Fully organised Media Pool and searchable transcripts
03
Claude AI · Python

Story Intelligence & AI Analysis

DocuFlow compresses your catalog and transcripts into a ~100–200 KB summary and sends it to Claude AI. The AI identifies the strongest soundbites, marks them in Resolve, generates story briefs from your footage, and produces multiple fully structured proposals — each with clip order, timecodes, track assignments, and editorial reasoning.

✓ 3–8 story proposals ready in ~2 minutes, ~$1–5 total cost
04
DaVinci Resolve · Lua

Build the Timeline

Run the timeline builder inside Resolve. Your chosen proposal becomes a fully structured rough cut: interviews on V1, b-roll on V2, tracks correctly routed. The first assembly is ready before the end of day one.

✓ Rough cut assembled in ~10 seconds. Edit from here.

Additional tools in the pipeline

🔍
Semantic Search Find any shot by describing it in plain English — "golden hour, contemplative mood"
🎬
Topic Mini-Cuts Generate a short cut on any theme in seconds — DocuFlow finds sync and b-roll, pairs them, sequences the edit
✂️
Silence Trimming Non-destructive removal of pauses and filler words using Whisper word-level timestamps
📊
Coverage Brief Pre-analysis readiness check — scores your catalog 0–100 for completeness before AI runs
Features

Everything a documentary editor needs, automated

🧠

AI Story Discovery

No brief required. DocuFlow reads your catalog, identifies themes, characters, and narrative arcs, and generates story briefs from your footage. Feed those briefs back in to produce multiple fully structured rough cut proposals.

🔍

Semantic Footage Search

Describe any shot in plain English. "Intimate close-up, contemplative mood", "someone mentioning funding struggles" — DocuFlow returns ranked results with suggested in/out timecodes and reasoning for each match.

🎯

AI Soundbite Identification

Claude reads every interview transcript and identifies the strongest soundbites — categorised as Blue (hero quote), Purple (strong moment), or Yellow (supporting). Markers appear in Resolve before you open the timeline.

Instant Rough Cut Assembly

Each AI proposal includes clip IDs, in/out timecodes, track assignments, and editorial reasoning. The timeline builder turns it into a Resolve timeline in seconds — interview on V1, b-roll on V2, audio routed correctly.

🔒

Fully Local — Footage Never Moves

Only a compressed ~100–200 KB metadata summary ever leaves your machine. No footage uploaded to any cloud. No third-party servers storing your client's material. Full privacy, full control.

📋

Multiple Story Versions

Generate 3, 5, or 8 distinct story proposals from the same footage in one run. Explore different angles, tones, and structures simultaneously — choose the strongest direction before committing to an edit.

🏷️

Editor Marker Integration

Your Blue, Purple, and Yellow markers in Resolve feed directly into the AI analysis. Proposals reflect your editorial judgment — moments you've already flagged as important carry extra weight in the AI's decisions.

🗂️

Persistent, Searchable Catalog

Every project builds a structured catalog that remains searchable years later. Find footage from past projects in seconds. The catalog is a permanent asset — not locked in someone's memory or an unlabelled drive.

🖥️

Local Web Interface

A clean browser-based UI lets non-technical team members run pipeline steps, browse results, and view proposals without touching the command line. One command starts it; a browser tab opens automatically.

Time Savings

Before and after, by the numbers

Every hour you spend on logistics is an hour you're not spending on the story. Here's what DocuFlow gives back.

Task Manual With DocuFlow Time saved
Classify + rename 200 clips 2–4 hours ~5 minutes 96% faster
Bin and organise in Resolve 1–2 hours ~1 minute 98% faster
Log transcripts 8–20 hours ~1 hour (AI) 90% faster
Identify key soundbites 4–8 hours ~30 seconds 99% faster
Story development + clip selection 1–3 days ~2 minutes Essentially instant
Build rough cut in timeline 2–4 hours ~10 seconds 99.9% faster
Find a specific shot 30–60 minutes ~15 seconds 99% faster
Remove pauses from a dialogue cut 1–2 hours ~30 seconds 99% faster
The pre-edit admin phase alone — classification, logging, story development, assembly — typically takes 3–5 days before a single creative decision is made. DocuFlow compresses it to a few hours.
Who It's For

Built for factual and documentary production

🎥

Documentary Editors

If you regularly receive 100–500 clips from a shoot and spend the first week just logging before you can begin editing, DocuFlow was built for you. Start editing on day one.

🏢

Corporate & Branded Content

Talking-head interviews, product b-roll, event coverage. DocuFlow reads every interview, pulls the strongest soundbites, and proposes structured cuts aligned to your brief — especially useful for multiple platform versions.

🎬

Directors Who Edit

You know the story you want to tell but don't want to spend days scrubbing footage. Describe the shot you're looking for and find it in seconds, or generate multiple story angles to explore before committing.

📡

Small Production Companies

Produce high volumes of factual content without a dedicated logging team. DocuFlow gives you the same organisational infrastructure that large post houses build with dedicated staff — automated, consistent, and reusable.

System Requirements

OS Windows 10 or 11 (64-bit)
NLE DaVinci Resolve 20 Studio (paid)
Python 3.10 or newer
API Anthropic API key (Claude AI)
RAM 16 GB min, 32 GB recommended
GPU NVIDIA GPU recommended for Whisper step

Mac / Linux support is not available yet — DocuFlow requires DaVinci Resolve's Lua scripting environment on Windows.

Pricing

Pay for what you use. Nothing more.

DocuFlow is self-hosted software. There is no monthly subscription. Your only ongoing cost is the Claude API calls your pipeline actually makes.

DocuFlow Software
Contact us
one-time licence
  • ✓ Full pipeline — ingest to timeline
  • ✓ Local web UI
  • ✓ All Lua scripts for DaVinci Resolve
  • ✓ All Python pipeline scripts
  • ✓ Unlimited projects
  • ✓ No monthly fee
Get in Touch
Claude API Usage
~$1–5
per full pipeline run
  • ✓ Billed directly by Anthropic
  • ✓ Pay only for what you run
  • ✓ Typical run: 3–5 proposals + search
  • ✓ Most steps run fully offline
  • ✓ Only metadata sent — never footage
  • ✓ ~100–200 KB per API call
Anthropic API →
Competitors charge $25–$100/month for tools that do a fraction of what DocuFlow does — and require uploading your footage to their servers.
How It Compares

The only tool that covers the full pipeline

Most AI video tools target podcasters and YouTubers, or do only one part of the job. DocuFlow is the only tool that covers the full documentary pipeline inside DaVinci Resolve, with footage staying on your machine.

Feature DocuFlow Eddie AI IntelliScript FireCut Descript
Runs inside DaVinci Resolve Export only
Footage stays local (no upload) ✗ cloud ✗ cloud
Ingest + bin classification
Story discovery from footage
Rough cut assembly Scripted only Partial
Semantic footage search
AI soundbite markers per clip
Silence / filler word removal
Built for 100–500 clip shoots Partial
Cost model API usage only $25–100/mo In Resolve Studio ~$24–34/mo $15–50/mo

Eddie AI is the closest competitor — it's well-built and worth knowing about. But it's a web app that requires footage upload, needs a story brief as input, and has no ingest pipeline or semantic search. DocuFlow generates the story brief from your footage, requires no input, and never leaves your machine.

FAQ

Frequently asked questions

How much time does DocuFlow actually save?

The short answer: DocuFlow turns a 2–3 week pre-edit workflow into a single day. In a standard documentary workflow, an editor spends roughly 80% of their time on logistics — logging footage, searching for shots, organising bins — and only 20% on actual creative decisions. DocuFlow automates the logistics. The rough cut isn't the end product — it's the starting point. What DocuFlow gives back is the time to refine it.

Does footage ever leave my machine?

No. DocuFlow never uploads footage anywhere. The AI steps work by sending a compressed metadata summary (~100–200 KB) to the Anthropic API — this contains transcript text, clip metadata, and marker information, but no video. Your footage stays on your local drive at all times. This is a deliberate design choice: documentary editors frequently work with sensitive or embargoed material.

Can I use DocuFlow with footage I've already ingested?

Yes. The catalog export step — which walks the Resolve Media Pool and builds master_catalog.json — works on whatever clips are already in Resolve. If your footage is already imported and organised into bins, you can skip the ingest step and start from catalog export. You can also reclassify clips at any time by moving them to the correct bin in Resolve and re-exporting the catalog — transcripts and markers are preserved.

What does the AI actually cost to run?

A full pipeline run — story intelligence, 5 proposals, a catalog search, and a topic mini-cut — typically costs $1–5 in Claude API usage, billed directly by Anthropic. The exact cost depends on your catalog size and how many AI steps you run. Most pipeline steps (ingest, import, transcription, catalog export, enrichment, silence trimming) run entirely offline with no API cost. Individual searches and topic cuts typically cost less than $0.50 each.

Do I need DaVinci Resolve Studio (paid)?

Yes. DocuFlow's Lua scripts use DaVinci Resolve's scripting API, which is only available in Resolve Studio (the paid version). The free version of Resolve does not expose the API that DocuFlow's scripts depend on. Resolve Studio is a one-time purchase (~$295) with no subscription.

How does DocuFlow compare to Eddie AI?

Eddie AI is the closest competitor — it produces rough cuts from interview footage and has a DaVinci Resolve export. But it's a web app: footage is uploaded to their servers, and you work outside Resolve. It also requires you to provide a story brief as input; DocuFlow generates the story brief from your footage with no input required. Eddie AI has no ingest pipeline, bin organisation, or semantic b-roll search. Pricing is $25–$100/month; DocuFlow's only ongoing cost is the Claude API calls you make.

Does DocuFlow work with non-documentary video?

DocuFlow is optimised for interview-driven and observational footage — documentaries, corporate content, event coverage, branded film. It works best when there's spoken word content to analyse. It is not currently optimised for narrative fiction (scripted drama), music videos, or purely visual/VFX-heavy work. The ingest, catalog, and search tools work for any footage type; the AI analysis and story discovery are most powerful with interview material.

Is this available for Mac or Linux?

Not yet. DocuFlow currently requires Windows due to how DaVinci Resolve's Lua scripting environment is accessed on Windows. Mac and Linux support is on the roadmap. Get in touch if this is important to you — it helps us prioritise.

Get in Touch

Ready to reclaim your editing time?

DocuFlow is currently in early access. Get in touch to learn more, ask technical questions, or discuss licensing for your team.

Quick responses We typically reply within one business day
🛠
Technical questions welcome Happy to discuss your specific workflow and footage volume
🔑
Early access pricing Founding customers receive reduced licensing rates

We'll only use your email to reply to your enquiry. No marketing, no mailing lists.