invideo: AI Video Editor

Review by eFreeware

Most AI video tools ask you to type a prompt and accept whatever comes back. invideo takes a different position: the AI works for you, and you keep the scissors. You tell Agent Two what you want made or changed in ordinary language, and the result lands on a multitrack timeline where every cut stays open to manual adjustment. That combination of direction and control is the most distinctive thing about it.

The app comes from invideo INC and sits in the Video Players & Editors category on Google Play. It is a free download with in-app purchases and no ads. It has passed 10 million installs and holds a 3.9 average from roughly 139,200 ratings, which suggests a large audience with mixed feelings rather than uniform praise.

I would point this at people who already think in edits: solo creators making reels and explainers, marketers building product ads, and small teams who need to review each other's work without exporting files back and forth. If you would rather never see a timeline, the hands-on layer will feel like more surface area than you need.

Directing with Agent Two instead of prompting

The headline feature is Agent Two, an agent you address in everyday language. It can build a video from a rough idea, work on one scene, or polish an existing cut. The developer frames the relationship as you directing and the agent editing, which is a meaningfully different pitch from one-shot prompt-to-video tools.

What I appreciate in that description is the loop it implies. You steer the narrative, rhythm, imagery and audio as the edit develops, then keep adjusting until it feels right. That is closer to working with an assistant editor than to spinning a slot machine, and it is the part of the app I would expect to matter most in practice.

The starting point is flexible. You can begin with an idea, with a script, or with your own footage. That last option is the one that separates invideo from generators that only produce synthetic clips: existing material can sit inside the same project as everything the models make.

A timeline that keeps AI edits editable

The second half of the pitch is the editor itself. Nothing the AI produces is locked, and manual tools sit on a multitrack timeline built for editors. You can arrange, trim and split clips, adjust layers, transitions, color and effects, use keyframes for animation, style titles and captions, and mix dialogue against music and sound effects.

That is a serious list for a mobile app, and it tells me the timeline is meant to be used, not just admired. The workflow the developer describes alternates between letting agents push the project forward and getting hands-on with the details, so you shape how each moment turns out.

For anyone who has tried to fix a generated clip in a tool with no real editing surface, the appeal here is obvious. The AI output is a draft you can rework, not a final answer you have to live with.

Generation across five media types

invideo bundles more than 200 AI models for image, video, voice, sound and music generation. In practice that means you can make a standalone image, animate a still, produce a video clip, turn written words into speech, or compose music for a new idea without leaving the app.

The developer's framing is about choice: pick the model that fits your creative direction and explore different possibilities. Having that many options is a genuine advantage when one model's output does not match the mood you are after, though it also means the browsing and selection layer carries a lot of weight.

Creations can be used on their own or brought into the editor alongside your own footage. That flexibility matters for people who sometimes need a single asset and sometimes a full production, and it keeps both jobs in one place.

Working with a team inside the project

Collaboration is built in rather than bolted on. Teams can join a project, work together live, and comment on individual frames. Feedback and editing remain linked as the work heads toward its final cut.

Frame-level comments are the detail I find most useful in that setup. Notes tied to an exact moment are far easier to act on than a general message about a rough cut, and they keep creative direction attached to the work instead of scattered across chat threads.

This is the feature that pushes invideo toward small production teams rather than lone creators. If two or three people need to agree on a cut before it ships, the review layer is likely to earn its place.

What it asks of you, and what it costs

The app is free to download and carries in-app purchases. It also states that there are no ads, which is a welcome detail for anyone who has edited inside an ad-supported tool. The free download is a starting point rather than a complete picture of cost.

The bigger requirement is conceptual. Agent Two works through language, and the timeline rewards people who understand clips, layers, keyframes and audio levels. That is not a barrier for editors, but it does mean the app asks you to engage with the craft rather than hand everything over.

There is also the matter of scale. A suite that spans generation, editing and review is doing a lot, and the 3.9 average across a very large number of ratings suggests the experience does not land equally well for everyone. I would go in expecting a tool that rewards patience with its structure.

Getting the most from the directing loop

The workflow the developer describes suggests a rhythm worth adopting: direct first, then edit. Let Agent Two build the rough shape of the piece, then move to the timeline for trimming, transitions, color and audio. Treating the agent as a first pass rather than a finisher is where this combination makes the most sense.

Scripts are worth preparing before you start, since the app accepts a script as an entry point alongside an idea or your own footage. A clear script gives the agent something concrete to work from and gives you a reference when judging the cut.

The generation models are best treated as a menu rather than a single button. When an image, voice or music result does not fit, trying a different model is the intended move, and the assets you keep can be dropped straight into the editor next to your own clips.

For team projects, I would set the review habit early. Frame-level feedback is most valuable when it arrives while the edit is still fluid, and it keeps the creative direction visible to everyone working on the final cut.

A timeline-first choice for creators who want to stay in control

The 3.9 average from about 139,200 ratings is the number I would weigh most heavily: respectable, but not a ringing endorsement.

The case for installing it rests on the pairing of Agent Two with a genuine multitrack editor. If you want to describe a change in plain language and then fix the result by hand, that loop is the reason to try it. The 200+ generation models and frame-level collaboration broaden the appeal to teams producing reels, explainers and short films.

I would recommend it to creators and small teams who are comfortable on a timeline and want AI to accelerate the first pass rather than replace their judgment. Anyone hoping for a single prompt that returns a finished video is likely to find the editing layer beside the point.

For everyone else, the sensible approach is to start with one project, direct it through Agent Two, and see how much of the final result you want to adjust yourself. That will tell you quickly whether the control this app offers is the control you actually want.

What's new in version 4.5.58

Meet the new invideo. - You direct, AI agents edit: tell Agent Two what to create or change in everyday language. - Take control of every cut on a professional multitrack timeline. - Generate images, video, voice, sound and music with 200+ AI models. - Collaborate in real time and leave feedback on specific frames.