Most AI video tools treat your footage like a one-way street.
You upload. It processes. It gives you an output. Done.
Story AI treats your footage differently. Once it runs through the pipeline, your material becomes a living corpus — something you can question, reshape, and direct. Not once, but continuously, as your understanding of the story evolves.
Here’s how it works, step by step.
Step 1: The Pipeline Runs on Your Footage
Before Story AI can do anything, your footage needs to be processed. That’s what the pipeline is for — a configurable chain of AI services that transform raw video into structured, machine-readable data.
You choose which stages run. You set the duration, pick the services, confirm the cost. The pipeline runs once — and everything it produces stays on the server, permanently linked to your project.
You can start with just Transcription. You can add stages later. The data accumulates over time, job by job, until your footage is as deeply analysed as your story needs it to be.
Changing your pipeline? Every new job reads your current saved configuration — so if you want different services next time, you need to reconfigure and activate the new setup before you run. Open Settings → Pipeline Configurator, adjust the services, hit Activate pipeline, and the next job will use the new config. How pipeline configuration works →
Step 2: Tell Story AI What You’re Making
Before processing begins, Story AI asks you a few questions. Not to fill out a form — to understand the edit you have in mind.
✦ Story AI — New Analysis
Answer a few questions and AI will structure your edit
These answers become the editorial brief that Story AI carries through the entire analysis. The same footage, briefed differently, produces a different story. A news package brief extracts hard facts and strong quotes. A documentary brief looks for emotional arcs and turning points. The AI is reading the same transcript — but through a different editorial lens.
Step 3: The Analysis Completes
Once processing finishes, Story AI produces a full structured analysis of your material:
You can read the full analysis, review the suggested edit plan, and then — this is where it gets interesting — keep talking to it.
Step 4: Chat With Your Transcript
The analysis doesn’t end when Story AI finishes. Below the result, there’s a chat interface directly connected to your processed material.
You can ask it anything about what it found:
Yes — three moments stand out.
At 00:08:42 he pauses mid-sentence: “We knew it was the right call, but… I’m not sure I slept that week.” The diarization shows a 2.3s silence before he continues — longer than anywhere else in the interview.
At 00:14:17 he uses the word “honestly” four times in 90 seconds, which in the sentiment model consistently flags as hedging.
At 00:31:05 he references his family directly for the first and only time. Emotion analysis rates it as the highest-valence segment in the file.
Here’s a proposed sequence for a 90-second emotional arc:
Opening: 00:08:38–00:09:02 — the pause and the sleepless admission
Mid: 00:14:10–00:14:55 — hedging and then the decision statement
Turn: 00:31:00–00:31:28 — family reference, most emotional segment
Close: 00:44:51–00:45:09 — “We didn’t go there to win. We went because we had to.”
The AI isn’t guessing. It’s reading the transcript, the emotion scores, the diarization data, and the visual analysis — all the data your pipeline already generated. Every answer comes with timecodes you can jump to directly.
Good questions to ask: “Find the moment the tone shifts.” · “Which speaker is most dominant in the first ten minutes?” · “Are there any sequences under 30 seconds that could work as a standalone clip?” · “What’s missing from the current edit plan that the transcript contains?”
Step 5: Turn a Chat Answer Into a Project
This is where the loop closes.
When you ask Story AI a question and check the “Create sandbox from answer” box, the question becomes something more than a query — it becomes a project creation prompt.
What “sandbox” means
Story AI creates a new project in your editor — pre-loaded with the clips, in/out points, and sequence order it proposed in the chat. It’s not a final edit. It’s a starting point you own.From here you work in AVScript as normal: adjust timecodes, reorder clips, change the pacing, export FCPXML to your NLE. The sandbox is yours. Story AI was the first editor.
The checkbox is the difference between asking “what could this look like?” and actually having it in your timeline.
The Loop, Not the Sequence
Most edit workflows are linear. You decide what you want, you build it, you export it, you’re done.
Story AI is a loop.
You process the footage. You read the analysis. You chat with the material. You discover something you didn’t plan for. You ask a new question. You build a new sandbox. You export a different version.
Meanwhile, the original processing — every transcript, every emotion score, every scene cut — is still on the server, stored on the media file itself, not just the project. The next time you open it, everything is still there. You haven’t lost anything. You’ve only added to it.
That distinction matters. If you delete a project, or rebuild your edit from scratch, the file carries all the enrichment with it. Add the same file to a new project and Story AI picks up exactly where it left off — no re-processing, no charge.
A documentary that took six months to shoot doesn’t need to be re-analysed every time the edit changes direction. The footage already knows what it contains. You just have to keep asking.
Where to Start
You don’t need to understand all of this before your first job. The short version:
- Pick your footage — any video file already in your library
- Set up a pipeline — Starter preset is enough to begin: Transcription + Story Summary
- Run Story AI — fill in the brief, hit Analyse
- Read what it found — the analysis panel shows the full breakdown
- Ask a question — type anything about your footage into the chat
- Check the box — if the answer looks like a good starting point, create the sandbox
The rest you’ll figure out once you’re talking to your footage.
→ Open Story AI · → Set up your pipeline · → Buy processing minutes