Video editing is powerful.
But it's still painfully manual.
Editors spend hours on repetitive tasks: finding the right clip, trimming B-roll, matching pacing, applying transitions. Current AI tools generate clips — but none of them truly understand what's in your footage or how to shape it into a story.
An agent that watches
before it edits.
AVA analyzes your footage frame by frame — understanding scenes, detecting action, recognizing speakers, and mapping the emotional arc of your content. This deep comprehension becomes the foundation for intelligent, context-aware editing.
The result: tell AVA what you want in plain language, and it builds the edit — not by guessing, but by understanding your material the way an experienced editor would.
You decide how much
the agent does.
AVA adapts to the scope of your request. A precise instruction gets a precise action. A broad creative brief gets an autonomous, multi-step workflow. The level of autonomy is always yours to choose.
Real prototype.
Working today.
Editing Inside Premiere
The agent works alongside your timeline — orchestrating subagents, calling tools, and editing clips in real time.
Chat Interface
You can instruct and observe ava throuch our Chat interface.
Project Media at a Glance
All project clips with extracted frames — select which to analyze or search.
Multi-Clip Analysis
Each clip is analyzed frame-by-frame — events, subjects, and compositions extracted automatically.
Scene-by-Scene Breakdown
Every action timestamped — the context the agent uses to make intelligent editing decisions.
The right tool surface
for every task.
Not every task belongs in the same window. AVA presents purpose-built interfaces that live where they make sense — inside your editor for timeline work, in dedicated panels for research and planning, seamlessly connected.
A real editing session
from prompt to finished (rough)cut.
3 tool calls completed
assets.searchByDescription
assets.searchByDescription
assets.searchByDescription
- 0:00–0:08 Kickflip trick — Clip 4
- 0:08–0:13 Reaction shot 1 — Clip 2
- 0:13–0:18 Reaction shot 2 — Clip 7
- 0:18–0:30 Wide sunset shot — Clip 16
6 tool calls completed
timeline.insertClip
timeline.insertClip
timeline.insertClip
timeline.insertClip
timeline.addTransition
timeline.addTransition
What we're building
where it leads.
Intelligent Editing
An AI agent inside your video editor that watches your footage, understands it, and edits through conversation.
Full Production Pipeline
Beyond editing — conceptualisation, writing, research, and gathering references, all feeding into the final cut.
Footage Intelligence
A fully indexed, semantically searchable footage database the agent draws from — across projects, teams, and editors.
Built by filmmakers
and engineers.
Anton von Heiseler Co-Founder
Masters student in Directing at Filmuniversität Babelsberg Konrad Wolf. Brings deep understanding of the creative process, narrative structure, and the real needs of filmmakers to AVA's product vision.
Johannes Wanner Co-Founder
Professional Software Engineer with a background from HTW Berlin. Specializes in distributed systems and AI infrastructure — building the backend that makes AVA's intelligent editing possible.
Jo Bettaque Technical Advisor
Professional Software Engineer from TU Berlin. Advises on architecture, AI/ML engineering, and system design — ensuring AVA's technology is built on solid foundations.
This is early-stage work.
We're looking for the right people.
Whether you're an investor, a potential integration partner, or a video professional who wants early access — we want to talk.
hello@ava-labs.de