Import and analyze
Bring in video, audio, or graphics. AVE builds useful project context from filenames, metadata, transcripts, visual notes, and timeline state.
How it works
From rough footage to approved timeline changes.
AVE separates creative reasoning from execution. AI helps understand the project and propose the path. Timeline changes still run through controlled tools you can inspect.
Ask or Plan
Use Ask for direction, search, and complete bounded edits. Use Plan when broader project-aware work should be previewed before execution.
Review real edits
AVE can create sequences, select clips, trim, add transitions, build captions, style graphics, and close gaps. Reversible actions stay undoable; broader plans remain reviewable before execution.
Export locally
Refine by hand, run gap-aware export preflight, and render finished deliverables at up to 4K UHD from the native Mac app.
Step 01
Start with messy footage, not a perfect prompt.
Import the footage you already have: a podcast recording, a launch shoot, generated clips, internal training material, or a folder of half-useful takes. AVE keeps the project local and builds context from the files, transcript, visual notes, metadata, and current timeline.
Local project folders
Imported media stays on your Mac by default
Transcript and visual context can guide planning
Step 02
Review each source before building the edit.
Select up to 50 imported videos and open Batch Selects. AVE inspects every recording separately for obvious image problems, abrupt movement, quiet sections, and shot boundaries. Preview and adjust the suggestions before accepting a separate selects sequence.
Per-recording review
Suggested removals and split points
Existing media and timelines stay unchanged
Step 03
Ask for a direction or request a plan.
Ask mode is for quick help and simple edits. Plan mode is for larger, project-aware work where AVE should preview the steps first. A good request can sound like a normal editor note, not a command-line recipe.
Ask for captions, search, trim help, or structure
Plan multi-step edits before execution
Use local models, installed assistants, or your own API setup
Step 04
Review the proposed changes before they touch the cut.
AVE shows the work as an edit plan: what it will search for, which sequence it will build, what format it will target, and which timeline actions it expects to run. You approve the path instead of trusting a hidden automation pass.
Approval-gated Plan mode
Visible tool calls
Accepted, declined, retried, and inspected changes
Step 05
Let AVE do the mechanical pass.
Once approved, supported actions run through AVE tools. That includes sequence creation, clip extraction, trims, splits, captions, transitions, format adaptation, styling, gap closing, and export preparation.
Real timeline mutations
Manual edits remain available
Native render and export preflight
Execution routing
Choose how much AI gets involved.
AVE gives you a routing choice because not every edit needs the same level of reasoning. Direct mechanical commands should feel instant. Editorial decisions can use the assistant you trust.
Auto
Best for normal work. AVE uses validated native tools when the request is clear, and asks your selected assistant when editorial reasoning is needed.
Fast
Best for direct commands. Fast keeps the request inside native editing actions for tasks like search, captions, splits, trims, and sequence setup.
Reason
Best for ambiguous creative choices. AVE asks the selected assistant first, then validates supported tool calls before changing the timeline.
Example requests
Talk like an editor, not a machine operator.
A good AVE request reads like a note to an editor. Each of these maps cleanly to a reviewable plan.
Review these recordings with Batch Selects before I build the cut.
Find the best quote about the launch, cut it vertical, and add captions.
Make a 45-second founder clip for LinkedIn from this interview.
Remove dead air, close the gaps, and keep the reaction shots.
Build a website hero cut and a 9:16 social version from the same footage.